The pool, cell by cell

pool snapshot 2026-09-18 · every row is a real community cell with its basis and run count · ranking provisional · a signed run outranks any claim

22 cells shown · 0 measured · 22 reported · MIN_RUNS_MEASURED = 3 · capture or correct a number ->

#modelrigbasisvaluenctxbits
1gpt-oss-20bchatM1 Max 64GBreported83.6 tok/s16408-bit
2gemma-4-26B-A4B-it-qat-GGUFchatM1 Max 64GBreported76.4 tok/s12,0484-bit
3gpt-oss-20b-GGUFchatM1 Max 64GBreported74.6 tok/s12,0484-bit
4Qwen3-Coder-30B-A3B-InstructcodeM1 Max 64GBreported72.3 tok/s12,0484-bit
5Qwen3-30B-A3BchatM1 Max 64GBreported62.6 tok/s12,0484-bit
6gpt-oss-20bchatM1 Max 64GBreported62.4 tok/s12,0484-bit
7Laguna-XS-2.1-GGUFchatM1 Max 64GBreported62.3 tok/s12,0484-bit
8Qwen3.6-35B-A3BchatM1 Max 64GBreported59.3 tok/s22,0484-bit
9Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16chatM1 Max 64GBreported57.9 tok/s17558-bit
10nvidia_Nemotron-Cascade-2-30B-A3B-GGUFchatM1 Max 64GBreported57.5 tok/s12,0484-bit
11Gemma-4-26B-A4B-NVFP4chatM1 Max 64GBreported52.7 tok/s12,0484-bit
12Llama-3.2-3B-InstructchatM1 Max 64GBreported50.9 tok/s164016-bit
13gemma-4-26B-A4B-itchatM1 Max 64GBreported49 tok/s22,0488-bit
14Qwen2.5-7B-InstructchatM1 Max 64GBreported48.7 tok/s12,0484-bit
15Mistral-Nemo-Instruct-2407chatM1 Max 64GBreported41.1 tok/s12,0484-bit
16Qwen3-14BchatM1 Max 64GBreported24.8 tok/s12,0484-bit
17gemma-4-12B-itchatM1 Max 64GBreported14 tok/s179416-bit
18Qwen3.6-27BchatM1 Max 64GBreported11.3 tok/s17878-bit
19Mistral-Small-3.2-24B-Instruct-2506chatM1 Max 64GBreported11.2 tok/s12,0484-bit
20Qwen3-32BchatM1 Max 64GBreported9 tok/s12,0484-bit
21gemma-4-31B-itchatM1 Max 64GBreported8.4 tok/s12,0488-bit
22BTL-3-CompactchatM1 Max 64GBreported3.8 tok/s22,0482-bit

Rows are community cells (medians), ranked provisionally. A signed run outranks any claim.