bestmodel.run / pool — agent view Honesty ladder: measured > reported > extrapolated > formula > no data yet. Top 60 cells, default sort: fastest decode. Columns: rig | model | basis | value | n. Filter: rig=rtx-3090-24gb. Full machine surface: /llms.txt · REST API: api.bestmodel.run rig | model | basis | value | n rtx-3090-24gb | liquidai-lfm2-5-1-2b-instruct-gguf | measured | 682.2 tok/s | n=4 rtx-3090-24gb | liquidai-lfm2-5-1-2b-instruct-gguf | measured | 564.4 tok/s | n=3 rtx-3090-24gb | baidu-unlimited-ocr | reported | 365.5 tok/s | n=1 rtx-3090-24gb | ggml-org-nvidia-nemotron-3-5-lightning-30b-a3b-gguf | reported | 353.7 tok/s | n=1 rtx-3090-24gb | liquidai-lfm2-5-2-6b-gguf | reported | 305.6 tok/s | n=1 rtx-3090-24gb | poolside-laguna-xs-2-1 | reported | 296 tok/s | n=1 rtx-3090-24gb | lmstudio-community-nvidia-nemotron-3-5-lightning-30b-a3b-gguf | reported | 202.6 tok/s | n=1 rtx-3090-24gb | unsloth-qwen3-6-35b-a3b-mtp-gguf | reported | 195.1 tok/s | n=1 rtx-3090-24gb | unsloth-qwen3-5-4b-mtp-gguf | reported | 191.2 tok/s | n=2 rtx-3090-24gb | bartowski-nvidia-nemotron-cascade-2-30b-a3b-gguf | measured | 190.3 tok/s | n=10 rtx-3090-24gb | nvidia-nemotron-cascade-2-30b-a3b | measured | 188.1 tok/s | n=9 rtx-3090-24gb | unsloth-gpt-oss-20b-gguf | reported | 180.4 tok/s | n=1 rtx-3090-24gb | arcee-ai-trinity-mini | measured | 162.6 tok/s | n=10 rtx-3090-24gb | openai-gpt-oss-20b | measured | 162.4 tok/s | n=9 rtx-3090-24gb | nvidia-nvidia-nemotron-3-nano-4b-bf16 | measured | 160.3 tok/s | n=10 rtx-3090-24gb | coherelabs-north-mini-code-1-0 | measured | 151.5 tok/s | n=10 rtx-3090-24gb | unsloth-qwen3-6-27b-gguf | reported | 150.1 tok/s | n=1 rtx-3090-24gb | migtissera-tess-4-35b-a3b | measured | 139.5 tok/s | n=12 rtx-3090-24gb | unsloth-gemma-4-26b-a4b-it-qat-gguf | measured | 138.3 tok/s | n=18 rtx-3090-24gb | unsloth-qwen3-5-9b-mtp-gguf | reported | 127.6 tok/s | n=2 rtx-3090-24gb | poolside-laguna-xs-2-1-gguf | measured | 125.2 tok/s | n=9 rtx-3090-24gb | unsloth-qwen3-5-35b-a3b-gguf | reported | 124.4 tok/s | n=1 rtx-3090-24gb | ornith-ai-ornith-1-0-35b-gguf | reported | 124.1 tok/s | n=1 rtx-3090-24gb | unsloth-qwen3-6-35b-a3b-mtp-gguf | reported | 121.6 tok/s | n=1 rtx-3090-24gb | unsloth-phi-4-mini-reasoning-gguf | measured | 121.6 tok/s | n=10 rtx-3090-24gb | mia-ailab-qwen3-8-27b-exl3-3-5bpw | measured | 115.1 tok/s | n=5 rtx-3090-24gb | qwen-qwen3-8-27b | reported | 113.2 tok/s | n=1 rtx-3090-24gb | poolside-laguna-xs-2-1-gguf | measured | 110.7 tok/s | n=9 rtx-3090-24gb | microsoft-phi-4-mini-reasoning | measured | 110.6 tok/s | n=9 rtx-3090-24gb | unsloth-gemma-4-12b-it-qat-gguf | measured | 104.1 tok/s | n=10 rtx-3090-24gb | unsloth-qwen3-6-27b-mtp-gguf | measured | 101.5 tok/s | n=91 rtx-3090-24gb | syvai-qwen3-8-27b-dflash2-w4a16 | reported | 101 tok/s | n=1 rtx-3090-24gb | tencent-hy-mt2-7b | reported | 98.1 tok/s | n=1 rtx-3090-24gb | prism-ml-ternary-bonsai-2-27b-gguf | reported | 95.7 tok/s | n=1 rtx-3090-24gb | unsloth-qwen3-6-35b-a3b | reported | 91.8 tok/s | n=1 rtx-3090-24gb | lucebox-deepseek-v4-flash-0731-rocmfp3 | measured | 78.9 tok/s | n=20 rtx-3090-24gb | empero-ai-qwythos-9b-v2-gguf | measured | 78.5 tok/s | n=10 rtx-3090-24gb | empero-ai-qwythos-9b-v2 | measured | 77.1 tok/s | n=9 rtx-3090-24gb | jackrong-deepseek-v4-pro-qwen3-5-9b-mtp-gguf | reported | 75.1 tok/s | n=2 rtx-3090-24gb | lorbus-qwen3-6-27b-int4-autoround | measured | 74.5 tok/s | n=4 rtx-3090-24gb | tencent-hunyuan-mt-7b | measured | 74.3 tok/s | n=9 rtx-3090-24gb | nvidia-nvidia-nemotron-nano-9b-v2 | measured | 73.7 tok/s | n=10 rtx-3090-24gb | unsloth-qwen3-8-27b-gguf | measured | 72.6 tok/s | n=6 rtx-3090-24gb | lmstudio-community-qwen3-6-27b-gguf | reported | 64.4 tok/s | n=1 rtx-3090-24gb | google-gemma-3-12b-it | measured | 59.6 tok/s | n=20 rtx-3090-24gb | huihui-ai-huihui-qwen3-6-27b-abliterated-mtp-gguf | reported | 57.2 tok/s | n=1 rtx-3090-24gb | ggml-org-gemma-4-12b-it-gguf | measured | 57 tok/s | n=4 rtx-3090-24gb | atomicchat-qwen3-6-27b-udt-mtp-gguf | reported | 56.2 tok/s | n=1 rtx-3090-24gb | ista-daslab-qwen3-8-27b-gsq-rco-gguf | reported | 56.1 tok/s | n=1 rtx-3090-24gb | nvidia-nvidia-nemotron-nano-12b-v2 | measured | 55.8 tok/s | n=10 rtx-3090-24gb | qwen-qwen3-8-27b | reported | 53 tok/s | n=1 rtx-3090-24gb | google-gemma-4-12b-it | measured | 52.3 tok/s | n=9 rtx-3090-24gb | unsloth-qwen3-8-27b-gguf | reported | 51.5 tok/s | n=1 rtx-3090-24gb | unsloth-qwen3-6-27b-mtp-gguf | measured | 49.2 tok/s | n=11 rtx-3090-24gb | nvidia-nemotron-cascade-14b-thinking | measured | 44.5 tok/s | n=10 rtx-3090-24gb | unsloth-qwen3-14b-gguf | measured | 44.4 tok/s | n=10 rtx-3090-24gb | xreyrobert-qwopus3-6-27b-v2-gptq-pro-v1 | reported | 43.2 tok/s | n=1 rtx-3090-24gb | bartowski-deepseek-r1-distill-qwen-14b-gguf | measured | 43 tok/s | n=10 rtx-3090-24gb | qwen-qwen3-14b | measured | 42.9 tok/s | n=9 rtx-3090-24gb | microsoft-phi-4-reasoning | measured | 41.2 tok/s | n=10 Capture or correct a number: /console (signed, Ed25519).