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...
60 cells shown · 24 measured · 36 reported · MIN_RUNS_MEASURED = 3 · capture or correct a number ->
| # | model | rig | basis | value | n | ctx | bits |
|---|---|---|---|---|---|---|---|
| 1 | Qwen3.6-27b-ROCmFP4-CIRU-CHAOS-SPEEDchat | Ryzen AI Max 395 128GB | reported | 132.3 tok/s | 1 | 2,048 | 4-bit |
| 2 | Qwen3.6-27Bchat | Ryzen AI Max 395 128GB | reported | 112 tok/s | 1 | 2,048 | 4-bit |
| 3 | gemma-4-E2B-itchat | Ryzen AI Max 395 128GB | measured | 107 tok/s | 3 | 32,768 | 4-bit |
| 4 | Qwen3-Coder-30B-A3B-Instructcode | Ryzen AI Max 395 128GB | measured | 80.4 tok/s | 5 | 100,000 | 5-bit |
| 5 | Qwen3.8-Flash-Next-GGUFchat | Ryzen AI Max 395 128GB | measured | 74.9 tok/s | 5 | 32,768 | 4-bit |
| 6 | Carnice-MoE-35B-A3Bchat | Ryzen AI Max 395 128GB | measured | 72.6 tok/s | 4 | 80,000 | 4-bit |
| 7 | Qwopus-MoE-35B-A3Bchat | Ryzen AI Max 395 128GB | measured | 72.5 tok/s | 6 | 128,000 | 4-bit |
| 8 | KAT-Coder-V2.5-Devcode | Ryzen AI Max 395 128GB | reported | 70.3 tok/s | 2 | 787 | 4-bit |
| 9 | Qwen3-Coder-30B-A3B-Instruct-GGUFcode | Ryzen AI Max 395 128GB | measured | 70.3 tok/s | 3 | 65,536 | 4-bit |
| 10 | Qwen3.8-27B-IU4-Kairic-Edgechat | Ryzen AI Max 395 128GB | measured | 66.9 tok/s | 3 | 32,768 | 4-bit |
| 11 | supergemma4-26b-uncensored-gguf-v2chat | Ryzen AI Max 395 128GB | reported | 66.1 tok/s | 1 | 16,384 | 4-bit |
| 12 | Qwen3.8-27Bchat | Ryzen AI Max 395 128GB | measured | 65 tok/s | 5 | 4,096 | 4-bit |
| 13 | Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressivechat | Ryzen AI Max 395 128GB | reported | 64.9 tok/s | 2 | 32,768 | 4-bit |
| 14 | Gemopus-4-26B-A4B-itchat | Ryzen AI Max 395 128GB | reported | 64.2 tok/s | 2 | 128,000 | 4-bit |
| 15 | Agents-A1-Q4_K_M-GGUFchat | Ryzen AI Max 395 128GB | reported | 62 tok/s | 1 | 262,144 | 4-bit |
| 16 | Qwen3.5-4Bchat | Ryzen AI Max 395 128GB | measured | 61.3 tok/s | 3 | 32,768 | 4-bit |
| 17 | gemma-4-26B-A4B-itchat | Ryzen AI Max 395 128GB | measured | 60.7 tok/s | 4 | 32,768 | 4-bit |
| 18 | Qwen3.6-35B-A3Bchat | Ryzen AI Max 395 128GB | measured | 57.5 tok/s | 5 | 131,072 | 4-bit |
| 19 | gemma-4-E4B-itchat | Ryzen AI Max 395 128GB | measured | 57.3 tok/s | 3 | 32,768 | 4-bit |
| 20 | Qwen3-Coder-Nextcode | Ryzen AI Max 395 128GB | measured | 55.8 tok/s | 4 | 80,000 | 4-bit |
| 21 | gemma-4-26B-A4B-itchat | Ryzen AI Max 395 128GB | reported | 55.5 tok/s | 1 | 16,384 | 6-bit |
| 22 | Qwopus-MoE-35B-A3B-GGUFchat | Ryzen AI Max 395 128GB | reported | 54.7 tok/s | 2 | 32,768 | 4-bit |
| 23 | Qwen3.6-35B-A3B-MTP-GGUFchat | Ryzen AI Max 395 128GB | measured | 54.5 tok/s | 8 | 262,144 | 4-bit |
| 24 | Ornith-1.0-35Bchat | Ryzen AI Max 395 128GB | reported | 54.1 tok/s | 1 | 787 | 8-bit |
| 25 | Qwable-3.6-35bchat | Ryzen AI Max 395 128GB | reported | 54.1 tok/s | 1 | 4,096 | 4-bit |
| 26 | Qwen3.8-Flash-Nextchat | Ryzen AI Max 395 128GB | reported | 54 tok/s | 1 | 262,144 | 4-bit |
| 27 | Qwopus-MoE-35B-A3Bchat | Ryzen AI Max 395 128GB | measured | 53.1 tok/s | 4 | 80,000 | 8-bit |
| 28 | Qwen3.6-35B-A3Bchat | Ryzen AI Max 395 128GB | measured | 52.8 tok/s | 3 | 128,000 | 8-bit |
| 29 | GLM-4.7-Flash-GGUFchat | Ryzen AI Max 395 128GB | reported | 52.8 tok/s | 2 | 4,096 | 4-bit |
| 30 | Qwen3.6-27b-ROCmFP4-CIRU-CHAOS-SPEEDchat | Ryzen AI Max 395 128GB | reported | 52.2 tok/s | 1 | 2,048 | 16-bit |
| 31 | Qwen3-Coder-Nextcode | Ryzen AI Max 395 128GB | measured | 51.2 tok/s | 5 | 80,000 | 5-bit |
| 32 | gemma-4-26B-A4B-itchat | Ryzen AI Max 395 128GB | reported | 46.1 tok/s | 2 | 16,384 | 8-bit |
| 33 | Gemopus-4-26B-A4B-itchat | Ryzen AI Max 395 128GB | reported | 45.8 tok/s | 2 | 128,000 | 8-bit |
| 34 | Ornith-1.0-35B-GGUFchat | Ryzen AI Max 395 128GB | reported | 43 tok/s | 2 | 65,536 | 4-bit |
| 35 | Qwen3.8-27B-ROCmFPX-GGUFchat | Ryzen AI Max 395 128GB | reported | 41 tok/s | 1 | 2,048 | 4-bit |
| 36 | Laguna-S-2.1chat | Ryzen AI Max 395 128GB | measured | 40.7 tok/s | 13 | 262,144 | 4-bit |
| 37 | Qwen3-8B-GGUFchat | Ryzen AI Max 395 128GB | reported | 39.3 tok/s | 1 | 4,096 | 4-bit |
| 38 | Ling-3.0-flashchat | Ryzen AI Max 395 128GB | reported | 39.2 tok/s | 1 | 4,096 | 4-bit |
| 39 | gpt-oss-120bchat | Ryzen AI Max 395 128GB | reported | 38.8 tok/s | 1 | 845 | 16-bit |
| 40 | Hermes-3-Llama-3.1-8Bchat | Ryzen AI Max 395 128GB | measured | 38.5 tok/s | 3 | 32,768 | 5-bit |
| 41 | Qwen3.5-9Bchat | Ryzen AI Max 395 128GB | measured | 37.2 tok/s | 3 | 32,768 | 4-bit |
| 42 | Qwythos-9B-Claude-Mythos-5-1M-GGUFchat | Ryzen AI Max 395 128GB | reported | 35.3 tok/s | 1 | 4,096 | 4-bit |
| 43 | Qwen-AgentWorld-35B-A3B-GGUFchat | Ryzen AI Max 395 128GB | reported | 34.9 tok/s | 2 | 65,536 | 4-bit |
| 44 | Bonsai-27B-ggufchat | Ryzen AI Max 395 128GB | measured | 33.6 tok/s | 4 | 262,144 | 1-bit |
| 45 | Ornith-1.0-9B-GGUFchat | Ryzen AI Max 395 128GB | reported | 31.8 tok/s | 1 | 4,096 | 5-bit |
| 46 | DeepSeek-V4-Flash-0731-ROCmFP3chat | Ryzen AI Max 395 128GB | reported | 30.1 tok/s | 2 | 8,192 | 2-bit |
| 47 | DeepSeek-V4-Flash-0731chat | Ryzen AI Max 395 128GB | reported | 28.3 tok/s | 2 | 782 | 2-bit |
| 48 | MiMo-V2.5chat | Ryzen AI Max 395 128GB | reported | 28.1 tok/s | 1 | 768 | 2-bit |
| 49 | Qwen3.8-27B-GGUFchat | Ryzen AI Max 395 128GB | measured | 26.7 tok/s | 5 | 131,072 | 4-bit |
| 50 | Gemma4-12B-QAT-Uncensored-HauhauCS-Balancedchat | Ryzen AI Max 395 128GB | reported | 24.8 tok/s | 1 | 4,096 | 4-bit |
| 51 | gemma-4-12B-coder-fable5-composer2.5-v1-GGUFcode | Ryzen AI Max 395 128GB | reported | 24.8 tok/s | 1 | 4,096 | 4-bit |
| 52 | gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUFchat | Ryzen AI Max 395 128GB | reported | 24.7 tok/s | 1 | 4,096 | 4-bit |
| 53 | Qwen3.8-27B-ROCmFP4-STRIX-MTP-GGUFchat | Ryzen AI Max 395 128GB | reported | 23.9 tok/s | 1 | 65,536 | 4-bit |
| 54 | Ternary-Bonsai-27B-ggufchat | Ryzen AI Max 395 128GB | measured | 23.2 tok/s | 4 | 262,144 | 2-bit |
| 55 | Qwythos-9B-Claude-Mythos-5-1M-GGUFchat | Ryzen AI Max 395 128GB | reported | 22.3 tok/s | 2 | 65,536 | 5-bit |
| 56 | Step-3.7-Flash-GGUFchat | Ryzen AI Max 395 128GB | reported | 22 tok/s | 1 | 65,536 | 4-bit |
| 57 | DeepSeek-V4-Flash-0731chat | Ryzen AI Max 395 128GB | reported | 21.8 tok/s | 1 | 782 | 1-bit |
| 58 | Qwen3.5-122B-A10Bchat | Ryzen AI Max 395 128GB | reported | 18 tok/s | 2 | 262,144 | 4-bit |
| 59 | Qwen3.6-27B-MTP-GGUFchat | Ryzen AI Max 395 128GB | measured | 17.1 tok/s | 3 | 65,536 | 4-bit |
| 60 | Qwen3.6-27Bchat | Ryzen AI Max 395 128GB | reported | 16.7 tok/s | 1 | 787 | 8-bit |
Rows are community cells (medians), ranked provisionally. A signed run outranks any claim.