{"meta":{"tiers":[{"tier":"T1","meaning":"reported \u2014 imported run file, tool+version pinned"},{"tier":"T0","meaning":"reported by submitters; no run file on record"}],"updated_at":"2026-10-31T00:48:36+00:00"},"boards":{"measured":[{"tier":"T1","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.47331427713900353637654916383326053619384765625,"ttft_ms":331.80228900164001970551908016204833984375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":11,"run_sha":"75865b17afa6237381a79212975f5e4aa1f24eff","run_file":"/api/v1/benchmarks/75865b17afa6237381a79212975f5e4aa1f24eff.json"},{"tier":"T1","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.7217847448689980183189618401229381561279296875,"ttft_ms":325.6752750021399833713076077401638031005859375,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":43.162181670367999686277471482753753662109375,"ttft_ms":314.40564501098998562156339175999164581298828125,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":10,"run_sha":"e06aa6e308c9ad9b9c4c66eac840631268c89dc0","run_file":"/api/v1/benchmarks/e06aa6e308c9ad9b9c4c66eac840631268c89dc0.json"},{"tier":"T1","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":43.162181670367999686277471482753753662109375,"ttft_ms":314.40564501098998562156339175999164581298828125,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.7217847448689980183189618401229381561279296875,"ttft_ms":325.6752750021399833713076077401638031005859375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":9,"run_sha":"01fc1673b2af058cee785ce1b8342f125bbe0628","run_file":"/api/v1/benchmarks/01fc1673b2af058cee785ce1b8342f125bbe0628.json"},{"tier":"T1","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.47331427713900353637654916383326053619384765625,"ttft_ms":331.80228900164001970551908016204833984375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":11,"run_sha":"75865b17afa6237381a79212975f5e4aa1f24eff","run_file":"/api/v1/benchmarks/75865b17afa6237381a79212975f5e4aa1f24eff.json"},{"tier":"T1","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":43.162181670367999686277471482753753662109375,"ttft_ms":314.40564501098998562156339175999164581298828125,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":10,"run_sha":"e06aa6e308c9ad9b9c4c66eac840631268c89dc0","run_file":"/api/v1/benchmarks/e06aa6e308c9ad9b9c4c66eac840631268c89dc0.json"},{"tier":"T1","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.7217847448689980183189618401229381561279296875,"ttft_ms":325.6752750021399833713076077401638031005859375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":9,"run_sha":"01fc1673b2af058cee785ce1b8342f125bbe0628","run_file":"/api/v1/benchmarks/01fc1673b2af058cee785ce1b8342f125bbe0628.json"},{"tier":"T1","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.7217847448689980183189618401229381561279296875,"ttft_ms":325.6752750021399833713076077401638031005859375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":9,"run_sha":"01fc1673b2af058cee785ce1b8342f125bbe0628","run_file":"/api/v1/benchmarks/01fc1673b2af058cee785ce1b8342f125bbe0628.json"},{"tier":"T1","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":43.162181670367999686277471482753753662109375,"ttft_ms":314.40564501098998562156339175999164581298828125,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":10,"run_sha":"e06aa6e308c9ad9b9c4c66eac840631268c89dc0","run_file":"/api/v1/benchmarks/e06aa6e308c9ad9b9c4c66eac840631268c89dc0.json"},{"tier":"T1","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.47331427713900353637654916383326053619384765625,"ttft_ms":331.80228900164001970551908016204833984375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":11,"run_sha":"75865b17afa6237381a79212975f5e4aa1f24eff","run_file":"/api/v1/benchmarks/75865b17afa6237381a79212975f5e4aa1f24eff.json"},{"tier":"T1","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.7217847448689980183189618401229381561279296875,"ttft_ms":325.6752750021399833713076077401638031005859375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":9,"run_sha":"01fc1673b2af058cee785ce1b8342f125bbe0628","run_file":"/api/v1/benchmarks/01fc1673b2af058cee785ce1b8342f125bbe0628.json"},{"tier":"T1","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.47331427713900353637654916383326053619384765625,"ttft_ms":331.80228900164001970551908016204833984375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":11,"run_sha":"75865b17afa6237381a79212975f5e4aa1f24eff","run_file":"/api/v1/benchmarks/75865b17afa6237381a79212975f5e4aa1f24eff.json"},{"tier":"T1","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":43.162181670367999686277471482753753662109375,"ttft_ms":314.40564501098998562156339175999164581298828125,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":10,"run_sha":"e06aa6e308c9ad9b9c4c66eac840631268c89dc0","run_file":"/api/v1/benchmarks/e06aa6e308c9ad9b9c4c66eac840631268c89dc0.json"},{"tier":"T1","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.7217847448689980183189618401229381561279296875,"ttft_ms":325.6752750021399833713076077401638031005859375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":9,"run_sha":"01fc1673b2af058cee785ce1b8342f125bbe0628","run_file":"/api/v1/benchmarks/01fc1673b2af058cee785ce1b8342f125bbe0628.json"},{"tier":"T1","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":43.162181670367999686277471482753753662109375,"ttft_ms":314.40564501098998562156339175999164581298828125,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":10,"run_sha":"e06aa6e308c9ad9b9c4c66eac840631268c89dc0","run_file":"/api/v1/benchmarks/e06aa6e308c9ad9b9c4c66eac840631268c89dc0.json"},{"tier":"T1","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","context_len":78,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":41.47331427713900353637654916383326053619384765625,"ttft_ms":331.80228900164001970551908016204833984375,"stddev":null,"slot_ceiling":4,"tool":"llm-inference-bench","tool_version":"0.8.0","method_ref":null,"measured_by":"key-2952","run_id":11,"run_sha":"75865b17afa6237381a79212975f5e4aa1f24eff","run_file":"/api/v1/benchmarks/75865b17afa6237381a79212975f5e4aa1f24eff.json"},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":null,"prefill_tok_s":6264,"decode_tok_s":null,"ttft_ms":null,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null},{"tier":"T1","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","context_len":262144,"concurrency":1,"prefill_tok_s":null,"decode_tok_s":264.3999999999999772626324556767940521240234375,"ttft_ms":63,"stddev":null,"slot_ceiling":null,"tool":null,"tool_version":null,"method_ref":null,"measured_by":null,"run_id":null,"run_sha":null,"run_file":null}],"claimed":[{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-cmp-170hx-tp2-dflash2-k7-hybrid","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"ExLlamaV3 @ 3a8e2fb84c + 9 patches (hash-guarded), sm_80 native","claimed_tok_s":124},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-cmp-170hx-tp2-dflash2-k7-hybrid","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"ExLlamaV3 @ 3a8e2fb84c + 9 patches (hash-guarded), sm_80 native","claimed_tok_s":124},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-cmp-170hx-tp2-dflash2-k7-hybrid","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"ExLlamaV3 @ 3a8e2fb84c + 9 patches (hash-guarded), sm_80 native","claimed_tok_s":124},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-cmp-170hx-tp2-dflash2-k7-hybrid","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"ExLlamaV3 @ 3a8e2fb84c + 9 patches (hash-guarded), sm_80 native","claimed_tok_s":124},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-cmp-170hx-dflash2-k7-q8-kv-384k","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"ExLlamaV3 1.5.4 + k_hcfuse (SM80)","claimed_tok_s":90},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-cmp-170hx-dflash2-k7-q8-kv-384k","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"ExLlamaV3 1.5.4 + k_hcfuse (SM80)","claimed_tok_s":90},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-cmp-170hx-dflash2-k7-q8-kv-384k","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"ExLlamaV3 1.5.4 + k_hcfuse (SM80)","claimed_tok_s":90},{"tier":"T0","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":43.16218167036803521341425948776304721832275390625},{"tier":"T0","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":43.16218167036803521341425948776304721832275390625},{"tier":"T0","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":43.16218167036803521341425948776304721832275390625},{"tier":"T0","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":43.16218167036803521341425948776304721832275390625},{"tier":"T0","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":43.16218167036803521341425948776304721832275390625},{"tier":"T0","recipe":"glm-5-3-flash-high-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":43.16218167036803521341425948776304721832275390625},{"tier":"T0","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.72178474486878485549823381006717681884765625},{"tier":"T0","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.72178474486878485549823381006717681884765625},{"tier":"T0","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.72178474486878485549823381006717681884765625},{"tier":"T0","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.72178474486878485549823381006717681884765625},{"tier":"T0","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.72178474486878485549823381006717681884765625},{"tier":"T0","recipe":"glm-5-3-flash-low-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.72178474486878485549823381006717681884765625},{"tier":"T0","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.4733142771386127378718811087310314178466796875},{"tier":"T0","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.4733142771386127378718811087310314178466796875},{"tier":"T0","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.4733142771386127378718811087310314178466796875},{"tier":"T0","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.4733142771386127378718811087310314178466796875},{"tier":"T0","recipe":"glm-5-3-flash-max-effort-on-2x-cmp170hx-exl3-tp2-four-slots","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"cmp-170hx-x2","engine":"exllamav3","engine_ver":"0f685aa + row32 four-slot profile","claimed_tok_s":41.4733142771386127378718811087310314178466796875},{"tier":"T0","recipe":"qwen3-8-flash-next-exl3-5-05bpw-2x-cmp-170hx-tp2-mtp3-morrowmake","model":"Qwen3.8-Flash-Next-EXL3-5.05bpw","quant":"exl3-5-05bpw","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"18305948ac (Morrowmake/vllm-cmp170hx)","claimed_tok_s":152.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-flash-next-exl3-5-05bpw-2x-cmp-170hx-tp2-mtp3-morrowmake","model":"Qwen3.8-Flash-Next-EXL3-5.05bpw","quant":"exl3-5-05bpw","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"18305948ac (Morrowmake/vllm-cmp170hx)","claimed_tok_s":152.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-awq-w4a16-on-2x-cmp-170hx-8-slots","model":"Qwen3.8-Flash-Next-Uncensored-AWQ","quant":"awq-w4a16-g32","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"wtdcode/vllm-backport @ 3142ded1 (source build) + recipe patches","claimed_tok_s":240.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-awq-w4a16-on-2x-cmp-170hx-8-slots","model":"Qwen3.8-Flash-Next-Uncensored-AWQ","quant":"awq-w4a16-g32","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"wtdcode/vllm-backport @ 3142ded1 (source build) + recipe patches","claimed_tok_s":192.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-awq-w4a16-on-2x-cmp-170hx-8-slots","model":"Qwen3.8-Flash-Next-Uncensored-AWQ","quant":"awq-w4a16-g32","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"wtdcode/vllm-backport @ 3142ded1 (source build) + recipe patches","claimed_tok_s":192.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-awq-w4a16-on-2x-cmp-170hx-8-slots","model":"Qwen3.8-Flash-Next-Uncensored-AWQ","quant":"awq-w4a16-g32","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"wtdcode/vllm-backport @ 3142ded1 (source build) + recipe patches","claimed_tok_s":192.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-awq-w4a16-on-2x-cmp-170hx-8-slots","model":"Qwen3.8-Flash-Next-Uncensored-AWQ","quant":"awq-w4a16-g32","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"wtdcode/vllm-backport @ 3142ded1 (source build) + recipe patches","claimed_tok_s":192.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-awq-w4a16-on-2x-cmp-170hx-8-slots","model":"Qwen3.8-Flash-Next-Uncensored-AWQ","quant":"awq-w4a16-g32","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"wtdcode/vllm-backport @ 3142ded1 (source build) + recipe patches","claimed_tok_s":192.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-awq-w4a16-on-2x-cmp-170hx-pp2-mtp3","model":"Qwen3.8-Flash-Next-Uncensored-AWQ","quant":"awq-w4a16-g32","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"wtdcode/vllm-backport @ 3142ded1 (source build)","claimed_tok_s":109},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-awq-w4a16-on-2x-cmp-170hx-pp2-mtp3","model":"Qwen3.8-Flash-Next-Uncensored-AWQ","quant":"awq-w4a16-g32","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"wtdcode/vllm-backport @ 3142ded1 (source build)","claimed_tok_s":109},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-2x-cmp-170hx-tp2-ep-mtp4-fp8-ple-92gb-h","model":"Qwen3.8-Flash-Next-W4A16","quant":"w4a16","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"qwen38-flash-next image + alesha-pro overlays @ 9b36aaec","claimed_tok_s":173},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-2x-cmp-170hx-tp2-ep-mtp4-fp8-ple-92gb-h","model":"Qwen3.8-Flash-Next-W4A16","quant":"w4a16","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"qwen38-flash-next image + alesha-pro overlays @ 9b36aaec","claimed_tok_s":173},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-2x-cmp-170hx-pp2-mtp5-ple-alexbrandes","model":"Qwen3.8-Flash-Next-W4A16-AutoRound","quant":"w4a16-autoround","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"vLLM v0.30.0 fork (02acf67c)","claimed_tok_s":226.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-2x-cmp-170hx-pp2-mtp5-ple-alexbrandes","model":"Qwen3.8-Flash-Next-W4A16-AutoRound","quant":"w4a16-autoround","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"vLLM v0.30.0 fork (02acf67c)","claimed_tok_s":226.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-2x-cmp-170hx-pp2-mtp5-ple-alexbrandes","model":"Qwen3.8-Flash-Next-W4A16-AutoRound","quant":"w4a16-autoround","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"vLLM v0.30.0 fork (02acf67c)","claimed_tok_s":226.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-fp8ple-2x-cmp-170hx-pp2-mtp4-1m","model":"Qwen3.8-Flash-Next-W4A16-FP8PLE","quant":"w4a16-fp8-ple","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"0.29 nightly (vllm==0.28.1rc1.dev93+cu133 verified)","claimed_tok_s":111.599999999999994315658113919198513031005859375},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-fp8ple-2x-cmp-170hx-pp2-mtp4-1m","model":"Qwen3.8-Flash-Next-W4A16-FP8PLE","quant":"w4a16-fp8-ple","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"0.29 nightly (vllm==0.28.1rc1.dev93+cu133 verified)","claimed_tok_s":111.599999999999994315658113919198513031005859375},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-2x-cmp-170hx-tp2-dflash2-host-ram-ple","model":"Qwen3.8-Flash-Next-W4A16-Merlin","quant":"w4a16-int4-experts-bf16-kv","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"fork @ 18305948acba (tag qwen38-recipe-1.0.0)","claimed_tok_s":253.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-int4ple-on-2x-modded-cmp-170hx-pp2-mtp3","model":"Qwen3.8-Flash-Next-W4A16-Merlin-INT4PLE-FP8KV","quant":"w4a16-int4-ple-int4","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"v0.30.0 @ ced6857a + 6-layer overlay (plugin zankich @ 6609bfef)","claimed_tok_s":133.3899999999999863575794734060764312744140625},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-int4ple-on-2x-modded-cmp-170hx-pp2-mtp3","model":"Qwen3.8-Flash-Next-W4A16-Merlin-INT4PLE-FP8KV","quant":"w4a16-int4-ple-int4","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"v0.30.0 @ ced6857a + 6-layer overlay (plugin zankich @ 6609bfef)","claimed_tok_s":133.3899999999999863575794734060764312744140625},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-int4ple-on-2x-modded-cmp-170hx-pp2-mtp3","model":"Qwen3.8-Flash-Next-W4A16-Merlin-INT4PLE-FP8KV","quant":"w4a16-int4-ple-int4","hardware":"cmp-170hx-x2","engine":"vllm","engine_ver":"v0.30.0 @ ced6857a + 6-layer overlay (plugin zankich @ 6609bfef)","claimed_tok_s":133.3899999999999863575794734060764312744140625},{"tier":"T0","recipe":"deepseek-v4-1-flash-sm80-deepinfer-kernel-port-6x-a100-dspark-k5","model":"DeepSeek-V4.1-Flash","quant":"fp8-official-checkpoint-fp8-latent-ue8m0-scales","hardware":"cmp-170hx-x6","engine":"vllm","engine_ver":"V1 + Deepinfer-SM80 patch @ 2a828fc2122b","claimed_tok_s":160.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"deepseek-v4-1-flash-sm80-deepinfer-kernel-port-6x-a100-dspark-k5","model":"DeepSeek-V4.1-Flash","quant":"fp8-official-checkpoint-fp8-latent-ue8m0-scales","hardware":"cmp-170hx-x6","engine":"vllm","engine_ver":"V1 @ vllm PR #56201 base 79a7108d9a + haosdent/vllm dsv4-flash-a100 SM80 base + Deepinfer-SM80 patch @ 8895510cae8f","claimed_tok_s":160.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"deepseek-v4-1-flash-sm80-deepinfer-kernel-port-6x-a100-dspark-k5","model":"DeepSeek-V4.1-Flash","quant":"fp8-official-checkpoint-fp8-latent-ue8m0-scales","hardware":"cmp-170hx-x6","engine":"vllm","engine_ver":"V1 + Deepinfer-SM80 patch @ 2a828fc2122b","claimed_tok_s":160.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-vq-q4-k-1x-dgx-spark-1m-ctx-jlfn","model":"GLM-5.3-Flash-VQ-Q4K","quant":"q4-k-vq-nonexpert-v3","hardware":"dgx-spark","engine":"other","engine_ver":"ds4-server (ds4-dfm-rs main @ c5384b5e+)","claimed_tok_s":24.300000000000000710542735760100185871124267578125},{"tier":"T0","recipe":"glm-5-3-flash-vq-q4-k-1x-dgx-spark-1m-ctx-jlfn","model":"GLM-5.3-Flash-VQ-Q4K","quant":"q4-k-vq-nonexpert-v3","hardware":"dgx-spark","engine":"other","engine_ver":"ds4-server (ds4-dfm-rs main @ c5384b5e+)","claimed_tok_s":24.300000000000000710542735760100185871124267578125},{"tier":"T0","recipe":"nemotron-3-5-bolt-74b-a9b-1x-dgx-spark-vllm-b12x-lilicorr-262k","model":"Nemotron-3.5-Bolt-74B-A9B-24GB","quant":"iq2-xxs-experts-nvfp4-mamba-q8-0-rest","hardware":"dgx-spark","engine":"vllm","engine_ver":"nightly-77154ca6 + b12x 1.5.1","claimed_tok_s":140},{"tier":"T0","recipe":"nemotron-3-5-bolt-74b-a9b-1x-dgx-spark-vllm-b12x-lilicorr-262k","model":"Nemotron-3.5-Bolt-74B-A9B-24GB","quant":"iq2-xxs-experts-nvfp4-mamba-q8-0-rest","hardware":"dgx-spark","engine":"vllm","engine_ver":"nightly-77154ca6 + b12x 1.5.1","claimed_tok_s":140},{"tier":"T0","recipe":"nemotron-3-5-bolt-74b-a9b-1x-dgx-spark-vllm-b12x-lilicorr-262k","model":"Nemotron-3.5-Bolt-74B-A9B-24GB","quant":"iq2-xxs-experts-nvfp4-mamba-q8-0-rest","hardware":"dgx-spark","engine":"vllm","engine_ver":"nightly-77154ca6 + b12x 1.5.1","claimed_tok_s":140},{"tier":"T0","recipe":"nemotron-3-5-bolt-74b-a9b-1x-dgx-spark-vllm-b12x-lilicorr-262k","model":"Nemotron-3.5-Bolt-74B-A9B-24GB","quant":"iq2-xxs-experts-nvfp4-mamba-q8-0-rest","hardware":"dgx-spark","engine":"vllm","engine_ver":"nightly-77154ca6 + b12x 1.5.1","claimed_tok_s":140},{"tier":"T0","recipe":"qwen3-8-27b-sglang-nvfp4-dflash2-on-1x-dgx-spark-deterministic","model":"Qwen3.8-27B","quant":"nvfp4","hardware":"dgx-spark","engine":"sglang","engine_ver":"v0.5.21","claimed_tok_s":71.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-27b-sglang-nvfp4-dflash2-on-1x-dgx-spark-deterministic","model":"Qwen3.8-27B","quant":"nvfp4","hardware":"dgx-spark","engine":"sglang","engine_ver":"v0.5.21","claimed_tok_s":71.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-27b-sglang-nvfp4-dflash2-on-1x-dgx-spark-deterministic","model":"Qwen3.8-27B","quant":"nvfp4","hardware":"dgx-spark","engine":"sglang","engine_ver":"v0.5.22","claimed_tok_s":68.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-fp4-local-1x-dgx-spark-vllm-b12x","model":"Qwen3.8-Flash-Next-FP4-Local","quant":"nvfp4","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev21830+g8244d1c44.d20261015","claimed_tok_s":66.1700000000000017053025658242404460906982421875},{"tier":"T0","recipe":"qwen3-8-flash-next-fp4-local-1x-dgx-spark-vllm-b12x","model":"Qwen3.8-Flash-Next-FP4-Local","quant":"nvfp4","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev21830+g8244d1c44.d20261015","claimed_tok_s":66.1700000000000017053025658242404460906982421875},{"tier":"T0","recipe":"qwen3-8-flash-next-fp4-local-1x-dgx-spark-vllm-b12x","model":"Qwen3.8-Flash-Next-FP4-Local","quant":"nvfp4","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev21830+g8244d1c44.d20261015","claimed_tok_s":66.1700000000000017053025658242404460906982421875},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-dgx-spark-gb10-sglang-nextn-mtp","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"dgx-spark","engine":"sglang","engine_ver":"image tag qwen38-flash:v1.5 (FROM lmsysorg/sglang:latest)","claimed_tok_s":110.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-dgx-spark-gb10-sglang-nextn-mtp","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"dgx-spark","engine":"sglang","engine_ver":"image tag qwen38-flash:v1.5 (FROM lmsysorg/sglang:latest)","claimed_tok_s":110.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-dgx-spark-gb10-sglang-nextn-mtp","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"dgx-spark","engine":"sglang","engine_ver":"image tag qwen38-flash:v1.5 (FROM lmsysorg/sglang:latest)","claimed_tok_s":110.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-1x-dgx-spark-vllm-71-3-c1-salient-data","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4-modelopt-dense-bf16-fp8-block-128","hardware":"dgx-spark","engine":"vllm","engine_ver":"nightly build 2026-10-14-77154ca6 (vLLM 77154ca6 + 14 patches)","claimed_tok_s":71.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-1x-dgx-spark-vllm-71-3-c1-salient-data","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4-modelopt-dense-bf16-fp8-block-128","hardware":"dgx-spark","engine":"vllm","engine_ver":"nightly build 2026-10-22-8e2beaa3 (vLLM 8e2beaa3 + 14 patches)","claimed_tok_s":69.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-1x-dgx-spark-vllm-71-3-c1-salient-data","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4-modelopt-dense-bf16-fp8-block-128","hardware":"dgx-spark","engine":"vllm","engine_ver":"nightly build 2026-10-22-8e2beaa3 (vLLM 8e2beaa3 + 14 patches)","claimed_tok_s":69.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-dgx-spark-gb10-sglang-mtp-213-262k","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"dgx-spark","engine":"sglang","engine_ver":"image commit d91c3682 (qwen4 source 73a255206f)","claimed_tok_s":32.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-dgx-spark-gb10-sglang-mtp-213-262k","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"dgx-spark","engine":"sglang","engine_ver":"image commit d91c3682 (qwen4 source 73a255206f)","claimed_tok_s":32.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-nvfp4-bf16-ple-dgx-spark-vllm-mtp2","model":"Qwen3.8-Flash-Next-Uncensored-NVFP4-BF16PLE","quant":"nvfp4","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev20073+g8e685d198 (official qwen38-flash-next image)","claimed_tok_s":44.22999999999999687361196265555918216705322265625},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-nvfp4-bf16-ple-dgx-spark-vllm-mtp2","model":"Qwen3.8-Flash-Next-Uncensored-NVFP4-BF16PLE","quant":"nvfp4","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev20073+g8e685d198 (official qwen38-flash-next image)","claimed_tok_s":44.22999999999999687361196265555918216705322265625},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-autoround-dgx-spark-vllm-mtp4-iter7g","model":"Qwen3.8-Flash-Next-W4A16-AutoRound-Hybrid","quant":"w4a16-autoround","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev20073 (qwen38-flash-next, iter7g floor)","claimed_tok_s":76.7999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-autoround-dgx-spark-vllm-mtp4-iter7g","model":"Qwen3.8-Flash-Next-W4A16-AutoRound-Hybrid","quant":"w4a16-autoround","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev20073 (qwen38-flash-next, iter7g floor)","claimed_tok_s":76.7999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-autoround-dgx-spark-vllm-mtp4-iter7g","model":"Qwen3.8-Flash-Next-W4A16-AutoRound-Hybrid","quant":"w4a16-autoround","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev20073 (qwen38-flash-next, iter7g floor)","claimed_tok_s":76.7999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-autoround-dgx-spark-vllm-mtp4-iter7g","model":"Qwen3.8-Flash-Next-W4A16-AutoRound-Hybrid","quant":"w4a16-autoround","hardware":"dgx-spark","engine":"vllm","engine_ver":"0.1.dev20073 (qwen38-flash-next, iter7g floor)","claimed_tok_s":76.7999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x2","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x2","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x2","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x2","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-nvfp4-2x-dgx-spark-sglang-tp2-roce","model":"Qwen3.8-Flash-Next-Uncensored-ModelOpt-NVFP4-FP8PLE","quant":"nvfp4","hardware":"dgx-spark-x2","engine":"sglang","engine_ver":null,"claimed_tok_s":117.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-nvfp4-2x-dgx-spark-sglang-tp2-roce","model":"Qwen3.8-Flash-Next-Uncensored-ModelOpt-NVFP4-FP8PLE","quant":"nvfp4","hardware":"dgx-spark-x2","engine":"sglang","engine_ver":null,"claimed_tok_s":117.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"qwen3-8-flash-next-uncensored-nvfp4-2x-dgx-spark-sglang-tp2-roce","model":"Qwen3.8-Flash-Next-Uncensored-ModelOpt-NVFP4-FP8PLE","quant":"nvfp4","hardware":"dgx-spark-x2","engine":"sglang","engine_ver":null,"claimed_tok_s":117.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x3","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x3","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x3","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x3","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"deepseek-v4-1-flash-on-4x-dgx-spark-tuned-tp4-sglang-dspark","model":"DeepSeek-V4.1-Flash","quant":"native-weights-mxfp4-experts-fp8-dense-deepseek-ai-release","hardware":"dgx-spark-x4","engine":"sglang","engine_ver":"dsv4.1 branch f80c91a4b + RoCEnante SG17 + adapters","claimed_tok_s":89.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"deepseek-v4-1-flash-on-4x-dgx-spark-tuned-tp4-sglang-dspark","model":"DeepSeek-V4.1-Flash","quant":"native-weights-mxfp4-experts-fp8-dense-deepseek-ai-release","hardware":"dgx-spark-x4","engine":"sglang","engine_ver":"dsv4.1 branch f80c91a4b + RoCEnante SG17 + adapters","claimed_tok_s":89.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"deepseek-v4-1-flash-on-4x-dgx-spark-tuned-tp4-sglang-dspark","model":"DeepSeek-V4.1-Flash","quant":"native-weights-mxfp4-experts-fp8-dense-deepseek-ai-release","hardware":"dgx-spark-x4","engine":"sglang","engine_ver":"dsv4.1 branch f80c91a4b + RoCEnante SG17 + adapters","claimed_tok_s":89.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x4","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x4","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x4","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-exlr8-exl3-experts-on-dgx-spark-tp1-4","model":"GLM-5.3-Flash-EXL3","quant":"exlr8-k2-k3-k4","hardware":"dgx-spark-x4","engine":"vllm","engine_ver":"vllm-exlr8@a2e163cb05","claimed_tok_s":53},{"tier":"T0","recipe":"glm-5-3-flash-nvfp4-4x-dgx-spark-switchless-ring-tp4-dflash2","model":"GLM-5.3-Flash-NVFP4","quant":"nvfp4-w4a16-modelopt-0-45","hardware":"dgx-spark-x4","engine":"vllm","engine_ver":"0.1.dev20051+g487ecf187","claimed_tok_s":71.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"glm-5-3-flash-nvfp4-4x-dgx-spark-switchless-ring-tp4-dflash2","model":"GLM-5.3-Flash-NVFP4","quant":"nvfp4-w4a16-modelopt-0-45","hardware":"dgx-spark-x4","engine":"vllm","engine_ver":"0.1.dev20051+g487ecf187","claimed_tok_s":71.2999999999999971578290569595992565155029296875},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-fp8-tp1xpp2-dspark-2x-dgx-station-gb300","model":"DeepSeek-V4.1-Flash-1017","quant":null,"hardware":"dgx-station-gb300-x2","engine":"vllm","engine_ver":"vLLM v0.31.0 + 5-file PP2+DSpark source overlay (bind-mount)","claimed_tok_s":391},{"tier":"T0","recipe":"qwen3-8-27b-int8-mtp3-prefix-cache-on-1x-cmp-170hx-hynix","model":"Qwen3.8-27B-INT8-W8A16-MTP","quant":"int8-w8a16","hardware":"nvidia-cmp-170hx","engine":"vllm","engine_ver":"0.26.1rc1.dev1024+g1fe3a1571 (upstream nightly wheel)","claimed_tok_s":65.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-27b-int8-mtp3-prefix-cache-on-1x-cmp-170hx-hynix","model":"Qwen3.8-27B-INT8-W8A16-MTP","quant":"int8-w8a16","hardware":"nvidia-cmp-170hx","engine":"vllm","engine_ver":"0.26.1rc1.dev1024+g1fe3a1571 (upstream nightly wheel)","claimed_tok_s":65.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-3bpw-fp8-ple-1x-cmp-170hx-dflash2-recoverssm","model":"Qwen3.8-Flash-Next-3bpw-AutoRound-MTP","quant":"ar-3bpw-fp8-ple-bf16-kv","hardware":"nvidia-cmp-170hx","engine":"vllm","engine_ver":"fork @ 18305948acba (tag qwen38-recipe-1.0.0)","claimed_tok_s":197.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-flash-next-3bpw-fp8-ple-1x-cmp-170hx-dflash2-recoverssm","model":"Qwen3.8-Flash-Next-3bpw-AutoRound-MTP","quant":"ar-3bpw-fp8-ple-bf16-kv","hardware":"nvidia-cmp-170hx","engine":"vllm","engine_ver":"fork @ 18305948acba (tag qwen38-recipe-1.0.0)","claimed_tok_s":197.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-flash-next-3bpw-mtp-on-1x-cmp-170hx-ssd-ple-aio-iiiusi0n","model":"Qwen3.8-Flash-Next-3bpw-AutoRound-MTP","quant":"auto-round-3bpw-mtp","hardware":"nvidia-cmp-170hx","engine":"vllm","engine_ver":"patched vLLM @ a5a30471 (ple-ssd patch set)","claimed_tok_s":111},{"tier":"T0","recipe":"ds4-flash-3x170hx-pp3-dspark-k5","model":"DeepSeek-V4-Flash-0731-abliterated","quant":"fp8-original","hardware":"nvidia-cmp-170hxx3","engine":"vllm","engine_ver":"patched (vllm-patched tree + dspark)","claimed_tok_s":181},{"tier":"T0","recipe":"ds4-flash-3x170hx-pp3-dspark-k5","model":"DeepSeek-V4-Flash-0731-abliterated","quant":"fp8-original","hardware":"nvidia-cmp-170hxx3","engine":"vllm","engine_ver":"patched (vllm-patched tree + dspark)","claimed_tok_s":181},{"tier":"T0","recipe":"ds4-flash-3x170hx-pp3-dspark-k5","model":"DeepSeek-V4-Flash-0731-abliterated","quant":"fp8-original","hardware":"nvidia-cmp-170hxx3","engine":"vllm","engine_ver":"patched (vllm-patched tree + dspark)","claimed_tok_s":181},{"tier":"T0","recipe":"ds4-flash-3x170hx-pp3-dspark-k5","model":"DeepSeek-V4-Flash-0731-abliterated","quant":"fp8-original","hardware":"nvidia-cmp-170hxx3","engine":"vllm","engine_ver":"patched (vllm-patched tree + dspark)","claimed_tok_s":181},{"tier":"T0","recipe":"deepseek-v4-1-flash-764b-on-4x-cmp-170hx-pp4-shadow-kv-exl3-2bpw","model":"DeepSeek-V4.1-Flash","quant":"exl3-2bpw","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"official vllm-openai:deepseekv41-flash-0909 + dsv41reap overlay","claimed_tok_s":54.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"deepseek-v4-1-flash-764b-on-4x-cmp-170hx-pp4-shadow-kv-exl3-2bpw","model":"DeepSeek-V4.1-Flash","quant":"exl3-2bpw","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"official vllm-openai:deepseekv41-flash-0909 + dsv41reap overlay","claimed_tok_s":54.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-fidelity-first-4x-cmp-170hx-alanoo81","model":"DeepSeek-V4.1-Flash-1017","quant":"exl3-hybrid-3bpw-all-40-layers","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"GNmac-lab recipe 1.2.0 @ 2ad5741ab1, patched","claimed_tok_s":105},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-engram-nvme-tp4-pp4-4x-cmp-170hx","model":"DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-Hybrid","quant":"exl3-2-9bpw-routed-experts-fp8-core-hybrid","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"patch scripts @ 2ad5741ab1 (built image, ~30-45 min)","claimed_tok_s":355.69999999999998863131622783839702606201171875},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-hybrid-exl3-tp4-4x-cmp-170hx-morrowmake","model":"DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-Hybrid","quant":"exl3-2-9bpw-routed-experts-fp8-core-hybrid","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"fork @ a0c57b52 (upstream 4e0681ed), image 1.0.0","claimed_tok_s":319},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-hybrid-exl3-tp4-4x-cmp-170hx-morrowmake","model":"DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-Hybrid","quant":"exl3-2-9bpw-routed-experts-fp8-core-hybrid","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"fork @ a0c57b52 (upstream 4e0681ed), image 1.0.0","claimed_tok_s":319},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-hybrid-exl3-tp4-4x-cmp-170hx-morrowmake","model":"DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-Hybrid","quant":"exl3-2-9bpw-routed-experts-fp8-core-hybrid","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"fork @ a0c57b52 (upstream 4e0681ed), image 1.0.0","claimed_tok_s":319},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-hybrid-exl3-tp4-4x-cmp-170hx-morrowmake","model":"DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-Hybrid","quant":"exl3-2-9bpw-routed-experts-fp8-core-hybrid","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"fork @ a0c57b52 (upstream 4e0681ed), image 1.0.0","claimed_tok_s":319},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-engram-nvme-tp4-pp4-4x-cmp-170hx","model":"DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-Hybrid","quant":"exl3-2-9bpw-routed-experts-fp8-core-hybrid","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"patch scripts @ bbaf367291 (built image, ~30-45 min)","claimed_tok_s":296},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-exl3-2-9bpw-tc-moe-on-4x-cmp-170hx-tp4","model":"DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-Hybrid","quant":"exl3-2-9bpw-routed-experts-fp8-core-hybrid","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"upstream vllm + Morrowmake sm80 host-allreduce 18305948","claimed_tok_s":180.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-exl3-2-9bpw-tc-moe-on-4x-cmp-170hx-tp4","model":"DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-Hybrid","quant":"exl3-2-9bpw-routed-experts-fp8-core-hybrid","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"upstream vllm + Morrowmake sm80 host-allreduce 18305948","claimed_tok_s":180.400000000000005684341886080801486968994140625},{"tier":"T0","recipe":"deepseek-v4-1-flash-exl3-3bpw-tc-moe-on-4x-cmp-170hx-tp4","model":"DeepSeek-V4.1-Flash-EXL3-3bpw","quant":"exl3-3bpw","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"upstream vllm + Morrowmake sm80 host-allreduce 18305948","claimed_tok_s":178.270000000000010231815394945442676544189453125},{"tier":"T0","recipe":"deepseek-v4-1-flash-exl3-3bpw-tc-moe-on-4x-cmp-170hx-tp4","model":"DeepSeek-V4.1-Flash-EXL3-3bpw","quant":"exl3-3bpw","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"upstream vllm + Morrowmake sm80 host-allreduce 18305948","claimed_tok_s":178.270000000000010231815394945442676544189453125},{"tier":"T0","recipe":"deepseek-v4-1-flash-exl3-3bpw-tc-moe-on-4x-cmp-170hx-tp4","model":"DeepSeek-V4.1-Flash-EXL3-3bpw","quant":"exl3-3bpw","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"upstream vllm + Morrowmake sm80 host-allreduce 18305948","claimed_tok_s":178.270000000000010231815394945442676544189453125},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 378c37b009","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ ab60b723ada","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ ab60b723ada","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ ab60b723ada","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ ab60b723ada","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ ab60b723ada","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"glm-5-3-flash-4x170hx-tp4-pp4-morrowmake-141","model":"GLM-5.3-Flash-W4A16-MTP","quant":"w4a16-int4-g128","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"Morrowmake/vllm-cmp170hx @ 819146023f","claimed_tok_s":264.3999999999999772626324556767940521240234375},{"tier":"T0","recipe":"mimo-v2-6-flash-rl-309b-moe-nvfp4-4x-cmp-170hx-tp4-dflash-1m-ctx","model":"MiMo-V2.6-Flash-RL","quant":"nvfp4","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"fork @ 18305948acba (tag qwen38-recipe-1.0.0)","claimed_tok_s":322},{"tier":"T0","recipe":"mimo-v2-6-flash-rl-309b-moe-nvfp4-4x-cmp-170hx-tp4-dflash-1m-ctx","model":"MiMo-V2.6-Flash-RL","quant":"nvfp4","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"fork @ 18305948acba (tag qwen38-recipe-1.0.0)","claimed_tok_s":322},{"tier":"T0","recipe":"mimo-v2-6-flash-rl-omni-on-4x-cmp-170hx-pp4-official-fp8-three-t","model":"MiMo-V2.6-Flash-RL","quant":"official-fp8","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"vllm-backport sm80 (v0.13.0 lineage), kaka86mm mimo-v3-stack","claimed_tok_s":64.5},{"tier":"T0","recipe":"mimo-v2-6-flash-rl-omni-on-4x-cmp-170hx-pp4-official-fp8-three-t","model":"MiMo-V2.6-Flash-RL","quant":"official-fp8","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"vllm-backport sm80 (v0.13.0 lineage), kaka86mm mimo-v3-stack","claimed_tok_s":64.5},{"tier":"T0","recipe":"qwen3-8-flash-next-fp8-4x-cmp-170hx-pp4-mtp-4-patched-vllm-g4bwy","model":"Qwen3.8-Flash-Next-FP8","quant":"fp8","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"mainline 183625d4f9 + 7-patch series","claimed_tok_s":212},{"tier":"T0","recipe":"qwen3-8-flash-next-fp8-4x-cmp-170hx-pp4-mtp-4-patched-vllm-g4bwy","model":"Qwen3.8-Flash-Next-FP8","quant":"fp8","hardware":"nvidia-cmp-170hxx4","engine":"vllm","engine_ver":"mainline 183625d4f9 + 7-patch series","claimed_tok_s":212},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-1x-rtx-pro-6000-gpu-expert-cache-h","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"exllamav3","engine_ver":"ExLlamaV3 1.6.0 + TabbyAPI 2fd6cc7 + expert cache (sm_120)","claimed_tok_s":61.47999999999999687361196265555918216705322265625},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-1x-rtx-pro-6000-gpu-expert-cache-h","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"exllamav3","engine_ver":"ExLlamaV3 1.6.0 + TabbyAPI 2fd6cc7 + expert cache (sm_120)","claimed_tok_s":61.47999999999999687361196265555918216705322265625},{"tier":"T0","recipe":"qwen3-8-27b-bit-exact-spec-decode-1x-rtx-pro-6000-sglang","model":"Qwen3.8-27B","quant":"bf16","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"main 542a043 (2026-09-23) + 2 patches + DeepGEMM 0.2.0 fence","claimed_tok_s":106.599999999999994315658113919198513031005859375},{"tier":"T0","recipe":"qwen3-8-27b-bit-exact-spec-decode-1x-rtx-pro-6000-sglang","model":"Qwen3.8-27B","quant":"bf16","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"main 542a043 (2026-09-23) + 2 patches + DeepGEMM 0.2.0 fence","claimed_tok_s":106.599999999999994315658113919198513031005859375},{"tier":"T0","recipe":"qwen3-8-27b-bit-exact-spec-decode-1x-rtx-pro-6000-sglang","model":"Qwen3.8-27B","quant":"bf16","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"main 542a043 (2026-09-23) + 2 patches + DeepGEMM 0.2.0 fence","claimed_tok_s":106.599999999999994315658113919198513031005859375},{"tier":"T0","recipe":"qwen3-8-27b-nvfp4a4-awq-1x-rtx-pro-6000-sglang-fast-decode","model":"Qwen3.8-27B-NVFP4A4-AWQ","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"0.5.21","claimed_tok_s":131.69999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-27b-nvfp4a4-awq-1x-rtx-pro-6000-sglang-fast-decode","model":"Qwen3.8-27B-NVFP4A4-AWQ","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"0.5.21","claimed_tok_s":131.69999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-27b-nvfp4a4-awq-1x-rtx-pro-6000-sglang-fast-decode","model":"Qwen3.8-27B-NVFP4A4-AWQ","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"0.5.22","claimed_tok_s":131.69999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-27b-nvfp4a4-awq-1x-rtx-pro-6000-sglang-fast-decode","model":"Qwen3.8-27B-NVFP4A4-AWQ","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"0.5.21","claimed_tok_s":131.69999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-27b-nvfp4a4-awq-1x-rtx-pro-6000-sglang-fast-decode","model":"Qwen3.8-27B-NVFP4A4-AWQ","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"0.5.21","claimed_tok_s":131.69999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-rtx-pro-6000-blackwell-sm120-sglang","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"0.5.19.dev series (repo fork)","claimed_tok_s":222.599999999999994315658113919198513031005859375},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-rtx-pro-6000-blackwell-sm120-sglang","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"0.5.19.dev series (repo fork)","claimed_tok_s":222.599999999999994315658113919198513031005859375},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-rtx-pro-6000-blackwell-sglang-nvme-ple","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"day-0 image + jzinno/sglang @ d4477bd298 + PRs 36556/36749/36750","claimed_tok_s":180.900000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-rtx-pro-6000-blackwell-sglang-nvme-ple","model":"Qwen3.8-Flash-Next-NVFP4","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"sglang","engine_ver":"day-0 image + jzinno/sglang @ d4477bd298 + PRs 36556/36749/36750","claimed_tok_s":180.900000000000005684341886080801486968994140625},{"tier":"T0","recipe":"qwen3-8-flash-next-nvfp4-1x6000-jj-mtp3","model":"Qwen3.8-Flash-Next-NVFP4","quant":"fp4-fp8-native","hardware":"nvidia-rtx-pro-6000-blackwell","engine":"vllm","engine_ver":"0.26.1rc0+glm53.r34 (Jovian Judgement community R34)","claimed_tok_s":117},{"tier":"T0","recipe":"qwen3-8-flash-next-1x6000-kk-mtp3","model":"Qwen3.8-Flash-Next","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-q","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":173.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-1x6000-kk-mtp3","model":"Qwen3.8-Flash-Next","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-q","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":173.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-1x6000-kk-mtp3","model":"Qwen3.8-Flash-Next","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-q","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":173.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-1x6000-kk-mtp3","model":"Qwen3.8-Flash-Next","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-q","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":173.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-1x6000-kk-mtp3","model":"Qwen3.8-Flash-Next","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-q","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":173.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"qwen3-8-flash-next-1x6000-kk-mtp3","model":"Qwen3.8-Flash-Next","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-q","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":173.19999999999998863131622783839702606201171875},{"tier":"T0","recipe":"glm-5-3-flash-qad-csf-on-2x-rtx-pro-6000-blackwell-tp2-dcp2-mtp3","model":"GLM-5.3-Flash","quant":"nvfp4-csf-mxfp8-attn-qad","hardware":"nvidia-rtx-pro-6000-blackwell-max-qx2","engine":"vllm","engine_ver":"karmic-kraken beta 20261003 (LIL community image)","claimed_tok_s":207},{"tier":"T0","recipe":"glm-5-3-flash-qad-csf-on-2x-rtx-pro-6000-blackwell-tp2-dcp2-mtp3","model":"GLM-5.3-Flash","quant":"nvfp4-csf-mxfp8-attn-qad","hardware":"nvidia-rtx-pro-6000-blackwell-max-qx2","engine":"vllm","engine_ver":"karmic-kraken beta 20261003 (LIL community image)","claimed_tok_s":207},{"tier":"T0","recipe":"glm-5-3-flash-4x6000-kk-dflash2-k7","model":"GLM-5.3-Flash","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-qx4","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":229},{"tier":"T0","recipe":"glm-5-3-flash-4x6000-kk-dflash2-k7","model":"GLM-5.3-Flash","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-qx4","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":229},{"tier":"T0","recipe":"glm-5-3-flash-4x6000-kk-dflash2-k7","model":"GLM-5.3-Flash","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-qx4","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":229},{"tier":"T0","recipe":"glm-5-3-flash-4x6000-kk-dflash2-k7","model":"GLM-5.3-Flash","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-qx4","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":229},{"tier":"T0","recipe":"glm-5-3-flash-4x6000-kk-dflash2-k7","model":"GLM-5.3-Flash","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-qx4","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":229},{"tier":"T0","recipe":"glm-5-3-flash-4x6000-kk-dflash2-k7","model":"GLM-5.3-Flash","quant":"nvfp4","hardware":"nvidia-rtx-pro-6000-blackwell-max-qx4","engine":"vllm","engine_ver":"karmic-kraken-beta (vLLM cfdecfaa4)","claimed_tok_s":229},{"tier":"T0","recipe":"qwen3-8-27b-mxfp4-dflash2-1x-radeon-ai-pro-r9700-wsl2-radiance","model":"Qwen3.8-27B-MXFP4","quant":"mxfp4-official","hardware":"radeon-ai-pro-r9700","engine":"vllm","engine_ver":"radiance 0.9.3 + ggz14/radiance-vllm-mxfp4 @ 7d9a15a","claimed_tok_s":125},{"tier":"T0","recipe":"qwen3-8-27b-mxfp4-dflash2-1x-radeon-ai-pro-r9700-wsl2-radiance","model":"Qwen3.8-27B-MXFP4","quant":"mxfp4-official","hardware":"radeon-ai-pro-r9700","engine":"vllm","engine_ver":"radiance 0.9.3 + ggz14/radiance-vllm-mxfp4 @ 45da506","claimed_tok_s":125},{"tier":"T0","recipe":"qwen3-8-27b-paro-mxfp6-2x-radeon-ai-pro-r9700-vllm-radiance-tp2","model":"Swift-Qwen3.8-27B-PARO-MXFP6","quant":"mxfp6-paro","hardware":"radeon-ai-pro-r9700-x2","engine":"vllm","engine_ver":"vllm-radiance 0.9.3 (vLLM v0.27.1 + RDNA4 patches)","claimed_tok_s":138},{"tier":"T0","recipe":"qwen3-8-27b-paro-mxfp6-2x-radeon-ai-pro-r9700-vllm-radiance-tp2","model":"Swift-Qwen3.8-27B-PARO-MXFP6","quant":"mxfp6-paro","hardware":"radeon-ai-pro-r9700-x2","engine":"vllm","engine_ver":"vllm-radiance 0.9.3 (vLLM v0.27.1 + RDNA4 patches)","claimed_tok_s":138},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-1x-rtx-3090-ram-nvme-expert-tierin","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"rtx-3090","engine":"exllamav3","engine_ver":"ExLlamaV3 1.5.1 (stock) + glm53 monkeypatch stack","claimed_tok_s":19.5},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-1x-rtx-3090-ram-nvme-expert-tierin","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"rtx-3090","engine":"exllamav3","engine_ver":"ExLlamaV3 1.5.1 (stock) + glm53 monkeypatch stack","claimed_tok_s":19.5},{"tier":"T0","recipe":"qwen3-8-27b-awq-int4-dspark-on-rtx-3090s-sglang-tp1-tp2-tp4","model":"Qwen3.8-27B-AWQ-INT4","quant":"awq-w4a16-g32","hardware":"rtx-3090","engine":"sglang","engine_ver":"git main @ 2e7c85da6 (torch 2.13.0+cu130)","claimed_tok_s":136.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-27b-awq-int4-dspark-on-rtx-3090s-sglang-tp1-tp2-tp4","model":"Qwen3.8-27B-AWQ-INT4","quant":"awq-w4a16-g32","hardware":"rtx-3090","engine":"sglang","engine_ver":"git main @ 2e7c85da6 (torch 2.13.0+cu130)","claimed_tok_s":136.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-fp8ple-on-1x-rtx-3090-cpu-cold-experts","model":"Qwen3.8-Flash-Next-W4A16-FP8PLE","quant":"w4a16-fp8-ple","hardware":"rtx-3090","engine":"vllm","engine_ver":"0.1.dev20073+g8e685d198 (qwen38-flash-next runtime)","claimed_tok_s":47.39999999999999857891452847979962825775146484375},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-fp8ple-on-1x-rtx-3090-cpu-cold-experts","model":"Qwen3.8-Flash-Next-W4A16-FP8PLE","quant":"w4a16-fp8-ple","hardware":"rtx-3090","engine":"vllm","engine_ver":"0.1.dev20073+g8e685d198 (qwen38-flash-next runtime)","claimed_tok_s":47.39999999999999857891452847979962825775146484375},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-dsx-2-6-bit-gptq-2x-rtx-3090-engram","model":"DeepSeek-V4.1-Flash-1017","quant":"dsx-mixed-2-6bit-per-expert-ptq1-0-q2-0-q3-k-gptq-dense-q8-0","hardware":"rtx-3090-x2","engine":"dsx","engine_ver":"dsx @ 98f1259732","claimed_tok_s":26},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-dsx-2-6-bit-gptq-2x-rtx-3090-engram","model":"DeepSeek-V4.1-Flash-1017","quant":"dsx-mixed-2-6bit-per-expert-ptq1-0-q2-0-q3-k-gptq-dense-q8-0","hardware":"rtx-3090-x2","engine":"dsx","engine_ver":"dsx 0.1.0 @ 971c072 (HEAD, unpinned upstream)","claimed_tok_s":26},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-rtx-3090-vllm-exllamav3-cpu-exp","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"rtx-3090-x2","engine":"vllm","engine_ver":"0.1.dev20073+g8e685d198 (qwen38-flash-next runtime)","claimed_tok_s":50.4200000000000017053025658242404460906982421875},{"tier":"T0","recipe":"glm-5-3-flash-exl3-3-05bpw-on-2x-rtx-3090-vllm-exllamav3-cpu-exp","model":"GLM-5.3-Flash-EXL3-3.05bpw","quant":"exl3-3-05bpw","hardware":"rtx-3090-x2","engine":"vllm","engine_ver":"0.1.dev20073+g8e685d198 (qwen38-flash-next runtime)","claimed_tok_s":50.4200000000000017053025658242404460906982421875},{"tier":"T0","recipe":"qwen3-8-27b-awq-int4-dspark-on-rtx-3090s-sglang-tp1-tp2-tp4","model":"Qwen3.8-27B-AWQ-INT4","quant":"awq-w4a16-g32","hardware":"rtx-3090-x2","engine":"sglang","engine_ver":"git main @ 2e7c85da6 (torch 2.13.0+cu130)","claimed_tok_s":136.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-27b-awq-int4-dspark-on-rtx-3090s-sglang-tp1-tp2-tp4","model":"Qwen3.8-27B-AWQ-INT4","quant":"awq-w4a16-g32","hardware":"rtx-3090-x2","engine":"sglang","engine_ver":"git main @ 2e7c85da6 (torch 2.13.0+cu130)","claimed_tok_s":136.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-fp8ple-on-2x-rtx-3090-tp2-expert-offloa","model":"Qwen3.8-Flash-Next-W4A16-FP8PLE","quant":"w4a16-fp8-ple","hardware":"rtx-3090-x2","engine":"vllm","engine_ver":"0.1.dev20073+g8e685d198 (qwen38-flash-next runtime)","claimed_tok_s":111.5},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-fp8ple-on-2x-rtx-3090-tp2-expert-offloa","model":"Qwen3.8-Flash-Next-W4A16-FP8PLE","quant":"w4a16-fp8-ple","hardware":"rtx-3090-x2","engine":"vllm","engine_ver":"0.1.dev20073+g8e685d198 (qwen38-flash-next runtime)","claimed_tok_s":111.5},{"tier":"T0","recipe":"qwen3-8-27b-awq-int4-dspark-on-rtx-3090s-sglang-tp1-tp2-tp4","model":"Qwen3.8-27B-AWQ-INT4","quant":"awq-w4a16-g32","hardware":"rtx-3090-x4","engine":"sglang","engine_ver":"git main @ 2e7c85da6 (torch 2.13.0+cu130)","claimed_tok_s":136.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-27b-awq-int4-dspark-on-rtx-3090s-sglang-tp1-tp2-tp4","model":"Qwen3.8-27B-AWQ-INT4","quant":"awq-w4a16-g32","hardware":"rtx-3090-x4","engine":"sglang","engine_ver":"git main @ 2e7c85da6 (torch 2.13.0+cu130)","claimed_tok_s":136.80000000000001136868377216160297393798828125},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-merlin-on-4x-rtx-3090-tp2xpp2-ep-mtp3-f","model":"Qwen3.8-Flash-Next-W4A16-Merlin","quant":"w4a16-fp8-kv","hardware":"rtx-3090-x4","engine":"vllm","engine_ver":"v0.30.0 + 97 commits (Flash-Next v2.5.1 build)","claimed_tok_s":533.700000000000045474735088646411895751953125},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-merlin-on-4x-rtx-3090-tp2xpp2-ep-mtp3-f","model":"Qwen3.8-Flash-Next-W4A16-Merlin","quant":"w4a16-fp8-kv","hardware":"rtx-3090-x4","engine":"vllm","engine_ver":"v0.30.0 + 97 commits (Flash-Next v2.5.1 build)","claimed_tok_s":533.700000000000045474735088646411895751953125},{"tier":"T0","recipe":"qwen3-8-flash-next-w4a16-merlin-on-4x-rtx-3090-tp2xpp2-ep-mtp3-f","model":"Qwen3.8-Flash-Next-W4A16-Merlin","quant":"w4a16-fp8-kv","hardware":"rtx-3090-x4","engine":"vllm","engine_ver":"v0.30.0 + 97 commits (Flash-Next v2.5.1 build)","claimed_tok_s":533.700000000000045474735088646411895751953125},{"tier":"T0","recipe":"glm-5-3-flash-exl3-2-05bpw-on-2x-rtx-5090-tabbyapi-hot-expert-sw","model":"GLM-5.3-Flash-EXL3-2.05bpw","quant":"exl3-2-05bpw","hardware":"rtx-5090-x2","engine":"exllamav3","engine_ver":"dev 5783a93 + TabbyAPI 53da7919","claimed_tok_s":64.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"glm-5-3-flash-exl3-2-05bpw-on-2x-rtx-5090-tabbyapi-hot-expert-sw","model":"GLM-5.3-Flash-EXL3-2.05bpw","quant":"exl3-2-05bpw","hardware":"rtx-5090-x2","engine":"exllamav3","engine_ver":"dev 5783a93 + TabbyAPI 53da7919","claimed_tok_s":64.7000000000000028421709430404007434844970703125},{"tier":"T0","recipe":"glm-5-3-flash-exl3-2-05bpw-on-2x-rtx-5090-tabbyapi-hot-expert-sw","model":"GLM-5.3-Flash-EXL3-2.05bpw","quant":"exl3-2-05bpw","hardware":"rtx-5090-x2","engine":"exllamav3","engine_ver":"dev 5783a93 + TabbyAPI 53da7919","claimed_tok_s":62.89999999999999857891452847979962825775146484375},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-kraken-tp3-3x-rtx-pro-6000-blackwell","model":"DeepSeek-V4.1-Flash-1017","quant":"native-weights-mxfp4-experts-fp8-dense-deepseek-ai-release","hardware":"rtx-pro-6000-blackwell-x3","engine":"vllm","engine_ver":"karmic-kraken-beta-20261016.2 + FlashInfer main 5b065abe","claimed_tok_s":186.69999999999998863131622783839702606201171875},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-kraken-tp3-3x-rtx-pro-6000-blackwell","model":"DeepSeek-V4.1-Flash-1017","quant":"native-weights-mxfp4-experts-fp8-dense-deepseek-ai-release","hardware":"rtx-pro-6000-blackwell-x3","engine":"vllm","engine_ver":"karmic-kraken-beta-20261016.2 + FlashInfer main 5b065abe","claimed_tok_s":186.69999999999998863131622783839702606201171875},{"tier":"T0","recipe":"deepseek-v4-1-flash-1017-kraken-tp3-3x-rtx-pro-6000-blackwell","model":"DeepSeek-V4.1-Flash-1017","quant":"native-weights-mxfp4-experts-fp8-dense-deepseek-ai-release","hardware":"rtx-pro-6000-blackwell-x3","engine":"vllm","engine_ver":"karmic-kraken-beta-20261016.2 + FlashInfer main 5b065abe","claimed_tok_s":186.69999999999998863131622783839702606201171875}]},"runs":[{"content_hash":"75865b17afa6237381a79212975f5e4aa1f24eff","tool":"llm-inference-bench","tool_version":"0.8.0","model":"GLM-5.3-Flash","engine":"exllamav3","engine_version":"0f685aa + row32 four-slot profile","deployment":{"layout_kind":"tp","layout_detail":"TP2 on 2x CMP170HX 64GiB, 4 active sequences","offload":"none"},"storage":["ram"],"tier":"spot-check","headline":{"decode_tok_s":41.4733142771386127378718811087310314178466796875},"created_at":"2026-10-17 12:50:29.696994-06","url":"/benchmarks/75865b17afa6237381a79212975f5e4aa1f24eff"},{"content_hash":"e06aa6e308c9ad9b9c4c66eac840631268c89dc0","tool":"llm-inference-bench","tool_version":"0.8.0","model":"GLM-5.3-Flash","engine":"exllamav3","engine_version":"0f685aa + row32 four-slot profile","deployment":{"layout_kind":"tp","layout_detail":"TP2 on 2x CMP170HX 64GiB, 4 active sequences","offload":"none"},"storage":["ram"],"tier":"spot-check","headline":{"decode_tok_s":43.16218167036803521341425948776304721832275390625},"created_at":"2026-10-17 12:49:49.373231-06","url":"/benchmarks/e06aa6e308c9ad9b9c4c66eac840631268c89dc0"},{"content_hash":"01fc1673b2af058cee785ce1b8342f125bbe0628","tool":"llm-inference-bench","tool_version":"0.8.0","model":"GLM-5.3-Flash","engine":"exllamav3","engine_version":"0f685aa + row32 four-slot profile","deployment":{"layout_kind":"tp","layout_detail":"TP2 on 2x CMP170HX 64GiB, 4 active sequences","offload":"none"},"storage":["ram"],"tier":"spot-check","headline":{"decode_tok_s":41.72178474486878485549823381006717681884765625},"created_at":"2026-10-17 12:49:09.575214-06","url":"/benchmarks/01fc1673b2af058cee785ce1b8342f125bbe0628"},{"content_hash":"74ae5d64b8eae6dcbdb14529d47d542f5b1459a7","tool":"llm-inference-bench","tool_version":"0.8.0","model":"GLM-5.3-Flash","engine":"exllamav3","engine_version":"0f685aa + row32 four-slot profile","deployment":{"layout_kind":"tp","layout_detail":"TP2 on 2x CMP170HX 64GiB, 4 active sequences","offload":"none"},"storage":["ram"],"tier":"spot-check","headline":{"decode_tok_s":null},"created_at":"2026-10-17 12:26:51.742549-06","url":"/benchmarks/74ae5d64b8eae6dcbdb14529d47d542f5b1459a7"},{"content_hash":"a6001d2b49a8d23fe6de6742d2e8ba2a6892f935","tool":"llm-inference-bench","tool_version":"0.8.0","model":"GLM-5.3-Flash","engine":"exllamav3","engine_version":"0f685aa + row32 four-slot profile","deployment":{"layout_kind":"tp","layout_detail":"TP2 on 2x CMP170HX 64GiB, 4 active sequences","offload":"none"},"storage":["ram"],"tier":"spot-check","headline":{"decode_tok_s":null},"created_at":"2026-10-17 12:26:03.095269-06","url":"/benchmarks/a6001d2b49a8d23fe6de6742d2e8ba2a6892f935"},{"content_hash":"4140365f3b3be8f4a92d60ea6fc957d3fbcd4872","tool":"llm-inference-bench","tool_version":"0.8.0","model":"GLM-5.3-Flash","engine":"exllamav3","engine_version":"0f685aa + row32 four-slot profile","deployment":{"layout_kind":"tp","layout_detail":"TP2 on 2x CMP170HX 64GiB, 4 active sequences","offload":"none"},"storage":["ram"],"tier":"spot-check","headline":{"decode_tok_s":null},"created_at":"2026-10-17 12:25:09.013243-06","url":"/benchmarks/4140365f3b3be8f4a92d60ea6fc957d3fbcd4872"}]}