Models
With measured speeds
- Llama 2 7B 6.74B
- Llama 3.2 1B Instruct 1.24B
- Llama 3.1 8B Instruct 8.03B
- Qwen2.5 14B Instruct 14.77B
- gpt-oss 20B 20.91B
- gpt-oss 120B 116.83B
- Qwen3 Coder 30B-A3B Instruct 30.53B
- Qwen2.5 Coder 7B 7.62B
- Gemma 3 4B IT (QAT) 3.88B
- GLM-4.7-Flash 29.94B
- Qwen3 4B Thinking 2507 4.02B
- Nemotron 3 Nano 4B 3.97B
- Nemotron 3 Super 120B-A12B 120.67B
Tracked from Hugging Face (fit and speed estimates)
- Spark-X2.6-4B 4.11B
- Agnes-3.0-Qwen 179.55B
- Ornith-1.5-122B 124.64B
- Tess-4.5-27B 27.32B
- Orbit-3.2-Nano 2.7B
- MLExpert 34.66B
- basal-omni-1.0-ultra 30.7B
- HybridAgent-14B-Agentic 14.77B
- Tessera-27B-B-CJK 27.32B
- MiniCPM-V-4.7-35B-A3B 34.66B
- Kolibri-1 78.1B
- DeepSeek-V4.1-Flash-1017 748.49B
- jebadiah-27b-v2-1 27.32B
- Beam-1.0-501B-A23B 501.27B
- NVIDIA-Nemotron-3.5-Super-VL-120B-A12B 120.0B
- NVIDIA-Nemotron-3.5-Bolt-74B-A9B 74.04B
- LFM2.5-24B-A2B 23.97B
- dwight-4b 4.02B
- d3 26.9B
- d3-nano 2.39B
- d3-mini 4.84B
- d3-lite 1.01B
- d3-flash 8.95B
- d3-edge 0.75B
- AesCode-32B 32.76B
- AesCode-8B 8.19B
- Mellum2.1-12B-A2.5B-Thinking 12.15B
- K2-Horizon-7B 9.0B
- MiMo-V2.6-Pro-RL 1021.25B
- clef-flash 9.08B
- clef 27.02B
- Darwin-27B-RSI 26.9B
- Artemis-31B-v1.2 30.7B
- MiMo-V2.6-Flash-RL 308.78B
- MiMo-V2.6-Distill-Qwen-9B 8.95B
- Hemmingway-1 27.32B
- Muse-Spark-Open-816B-A42B 819.99B
- Orion-26B-A4B-v1.1 25.23B
- Nex-N2.5-mini 34.66B
- DeepSeek-V4-Flash-Vision-Exp 284.33B
- GLM-5.3 753.86B
- GLM-5.3-Flash 320.76B
- Qwen3.8-Flash-Next 176.94B
- GLM-4.5-Air 106.85B
- GLM-4.5V 106.85B
- Qwen3.8-27B 27.32B
- Muse-Glimmer-30B 27.85B
- Nanbeige4.2-3B 4.17B
- Qwen3-8B-Base 8.19B
- gemma-4-31B 30.7B
- Qwen3-0.6B-Base 0.6B
- gemma-4-E2B 4.65B
- Gemma-4-31B-StyleTune 32.11B
- gemma-4-12B-it-qat-q4_0-unquantized 11.91B
- gemma-4-26B-A4B-it-qat-q4_0-unquantized 25.23B
- gemma-4-31B-it-qat-q4_0-unquantized 30.7B
- gemma-4-12B-it 11.91B
- LFM2.5-8B-A1B 8.47B
- Qwen3.5-0.8B 0.77B
- Qwen3.6-35B-A3B 34.66B
- Qwen3-ASR-1.7B 2.03B
- gemma-4-26B-A4B-it 25.23B
- gemma-4-E4B-it 7.52B
- gemma-4-31B-it 30.7B
- gemma-4-E2B-it 4.65B
- Qwen3.5-4B 4.21B
- Qwen3.5-9B 8.95B
- GLM-OCR 0.89B
- Kimi-Linear-48B-A3B-Instruct 49.12B
- Qwen3-Coder-Next 79.67B
- Qwen3-VL-30B-A3B-Instruct 30.53B
- granite-docling-258M 0.16B
- SmolLM3-3B 3.08B
- Mistral-Small-3.2-24B-Instruct-2506 23.57B
- DeepSeek-R1-0528-Qwen3-8B 8.19B
- Qwen3-8B 8.19B
- Qwen3-0.6B 0.75B
- Qwen3-30B-A3B 30.53B
- Qwen3-14B 14.77B
- Qwen3-32B 32.76B
- Qwen3-4B 4.02B
- gemma-3-12b-it 11.77B
- gemma-3-27b-it 27.01B
- gemma-3-4b-it 3.88B
- gemma-3-1b-it 1.0B
- Phi-4-mini-instruct 3.84B
- huihui-ai_Mistral-Small-24B-Instruct-2501-abliterated 23.57B
- DeepSeek-R1-Distill-Qwen-32B 32.76B
- DeepSeek-R1-Distill-Qwen-14B 14.77B
- Llama-3.3-70B-Instruct 70.55B
- Qwen2.5-Coder-7B-Instruct-abliterated 7.62B
- Qwen2.5-Coder-32B-Instruct 32.76B
- Llama-3.2-3B-Instruct-uncensored 3.61B
- Llama-3.2-3B-Instruct 3.21B
- Mistral-7B-Instruct-v0.3 7.25B