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
reading the pool...
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
reading the pool...
28 cells shown · 7 measured · 21 reported · MIN_RUNS_MEASURED = 3 · capture or correct a number ->
| # | model | rig | basis | value | n | ctx | bits |
|---|---|---|---|---|---|---|---|
| 1 | LFM2.5-8B-A1B-GGUFchat | RTX 3080 12GB | reported | 328.4 tok/s | 1 | 128,000 | 4-bit |
| 2 | LFM2.5-8B-A1B-GGUFchat | RTX 3080 12GB | reported | 277.8 tok/s | 1 | 128,000 | 6-bit |
| 3 | DeepSeek-R1-Distill-Qwen-7Bchat | RTX 3080 12GB | reported | 144 tok/s | 1 | 2,048 | 4-bit |
| 4 | Qwen2.5-Coder-7B-Instructcode | RTX 3080 12GB | measured | 140.8 tok/s | 3 | 2,048 | 4-bit |
| 5 | Llama-3.1-8B-Instructchat | RTX 3080 12GB | reported | 134 tok/s | 2 | 2,048 | 4-bit |
| 6 | gemma-4-E4B-itchat | RTX 3080 12GB | measured | 130 tok/s | 3 | 2,048 | 4-bit |
| 7 | NVIDIA-Nemotron-Nano-9B-v2chat | RTX 3080 12GB | reported | 98 tok/s | 1 | 2,048 | 4-bit |
| 8 | Ornith-1.5-9B-GGUFchat | RTX 3080 12GB | reported | 92 tok/s | 1 | 262,144 | 5-bit |
| 9 | DeepSeek-Coder-V2-Lite-Instructcode | RTX 3080 12GB | reported | 87 tok/s | 2 | 2,048 | 4-bit |
| 10 | Ornith-1.5-9B-GGUFchat | RTX 3080 12GB | reported | 82.8 tok/s | 1 | 131,072 | 6-bit |
| 11 | phi-4chat | RTX 3080 12GB | reported | 77 tok/s | 1 | 2,048 | 4-bit |
| 12 | gemma-3n-E4B-itchat | RTX 3080 12GB | reported | 75 tok/s | 2 | 2,048 | 4-bit |
| 13 | Qwen3-14Bchat | RTX 3080 12GB | reported | 75 tok/s | 1 | 2,048 | 4-bit |
| 14 | Bonsai-27B-ggufchat | RTX 3080 12GB | reported | 72.9 tok/s | 1 | 32,768 | 1-bit |
| 15 | Qwen2.5-Coder-14B-Instructcode | RTX 3080 12GB | measured | 72 tok/s | 5 | 2,048 | 4-bit |
| 16 | gemma-4-12B-it-qat-q4_0-ggufchat | RTX 3080 12GB | reported | 71.2 tok/s | 2 | 262,144 | 4-bit |
| 17 | Ornith-1.5-35B-A3B-GGUFchat | RTX 3080 12GB | reported | 60.8 tok/s | 2 | 131,072 | 4-bit |
| 18 | Qwen3.8-27B-Uncensored-GGUFchat | RTX 3080 12GB | reported | 60.7 tok/s | 2 | 16,384 | 2-bit |
| 19 | DeepCoder-14B-Previewcode | RTX 3080 12GB | reported | 59 tok/s | 1 | 2,048 | 4-bit |
| 20 | gpt-oss-20b-GGUFchat | RTX 3080 12GB | reported | 57.1 tok/s | 1 | 32,768 | 4-bit |
| 21 | gemma-3-12b-itchat | RTX 3080 12GB | measured | 51.5 tok/s | 4 | 2,048 | 4-bit |
| 22 | Qwen3.8-27B-GGUFchat | RTX 3080 12GB | measured | 44.4 tok/s | 3 | 65,536 | 2-bit |
| 23 | Qwen2.5-14B-Instructchat | RTX 3080 12GB | measured | 44 tok/s | 4 | 2,048 | 4-bit |
| 24 | Qwen3.6-35B-A3Bchat | RTX 3080 12GB | measured | 35.4 tok/s | 5 | 196,608 | 4-bit |
| 25 | Qwen3-30B-A3Bchat | RTX 3080 12GB | reported | 31 tok/s | 1 | 2,048 | 4-bit |
| 26 | gpt-oss-20bchat | RTX 3080 12GB | reported | 21 tok/s | 1 | 2,048 | 4-bit |
| 27 | Muse-Glimmer-30B-GGUFchat | RTX 3080 12GB | reported | 13.8 tok/s | 1 | 16,384 | 3-bit |
| 28 | Muse-Glimmer-30B-GGUFchat | RTX 3080 12GB | reported | 6 tok/s | 1 | 16,384 | 4-bit |
Rows are community cells (medians), ranked provisionally. A signed run outranks any claim.