Models
138 open models from 18 makers, with their published architecture. Pick one to see memory per format, KV cache per user and which GPUs it fits on.
138 models
| DeepSeek V4.1 FlashDeepSeek · DeepSeek V4 | 763B · 16.0B act. | 1,024K | MQA | 1,422 GB | 713 GB | 381 GB | 2026-09 |
| Qwen3.8 27BAlibaba Qwen · Qwen3.8 | 27.8B | 256K | hybrid | 51.7 GB | 30.6 GB | 20.7 GB | 2026-08 |
| Qwen3.8 2.4T-A95BAlibaba Qwen · Qwen3.8 | 2,446B · 95.3B act. | 256K | hybrid | 4,556 GB | 2,286 GB | 1,221 GB | 2026-08 |
| GLM-5.3Zhipu AI (Z.ai) · GLM-5 | 753B · 40.0B act. | 1,024K | MLA | 1,403 GB | 705 GB | 378 GB | 2026-08 |
| GLM-5.3-FlashZhipu AI (Z.ai) · GLM-5 | 321B · 18.0B act. | 1,024K | hybrid | 599 GB | 302 GB | 162 GB | 2026-08 |
| Nemotron 3.5 Lightning 30B-A3BNVIDIA · Nemotron 3 | 31.6B · 3.00B act. | 1,024K | hybrid | 58.8 GB | 30.7 GB | 17.6 GB | 2026-08 |
| Qwen3.8-Flash-Next 180B-A6BAlibaba Qwen · Qwen3.8 | 180B · 6.00B act. | 256K | hybrid | 335 GB | 170 GB | 92.5 GB | 2026-08 |
| Kimi K3Moonshot AI · Kimi K3 | 2,780B · 104B act. | 1,024K | hybrid | 5,178 GB | 2,593 GB | 1,382 GB | 2026-07 |
| Gemma 4 12BGoogle · Gemma 4 | 12.0B | 256K | GQA | 22.3 GB | 12.1 GB | 7.3 GB | 2026-06 |
| GLM-5.2Zhipu AI (Z.ai) · GLM-5 | 753B · 40.0B act. | 1,024K | MLA | 1,403 GB | 705 GB | 378 GB | 2026-06 |
| MiniMax-M3MiniMax · MiniMax | 427B · 23.0B act. | 1,024K | GQA | 795 GB | 402 GB | 218 GB | 2026-06 |
| Nemotron 3 Ultra 550B-A55BNVIDIA · Nemotron 3 | 561B · 55.0B act. | 1,024K | hybrid | 1,044 GB | 526 GB | 283 GB | 2026-06 |
| Command A+ (05-2026)Cohere · Command | 219B · 25.0B act. | 128K | GQA | 407 GB | 205 GB | 110 GB | 2026-05 |
| Qwen3.6 35B-A3BAlibaba Qwen · Qwen3.6 | 36.0B · 3.45B act. | 256K | hybrid | 67.0 GB | 35.4 GB | 20.6 GB | 2026-04 |
| Qwen3.6 27BAlibaba Qwen · Qwen3.6 | 27.8B | 256K | hybrid | 51.7 GB | 30.6 GB | 20.7 GB | 2026-04 |
| DeepSeek V4 ProDeepSeek · DeepSeek V4 | 1,599B · 49.0B act. | 1,024K | MQA | 2,978 GB | 1,492 GB | 796 GB | 2026-04 |
| DeepSeek V4 FlashDeepSeek · DeepSeek V4 | 291B · 13.0B act. | 1,024K | MQA | 542 GB | 273 GB | 147 GB | 2026-04 |
| Mistral Medium 3.5 128BMistral AI · Mistral Medium | 128B | 256K | GQA | 238 GB | 125 GB | 72.0 GB | 2026-04 |
| Kimi K2.6Moonshot AI · Kimi K2 | 1,027B · 32.9B act. | 256K | MLA | 1,913 GB | 961 GB | 514 GB | 2026-04 |
| MiniMax-M2.7MiniMax · MiniMax | 229B · 11.0B act. | 200K | GQA | 426 GB | 215 GB | 117 GB | 2026-04 |
| Granite 4.1 3BIBM · Granite 4.1 | 3.40B | 128K | GQA | 6.3 GB | 3.4 GB | 2.0 GB | 2026-04 |
| Granite 4.1 8BIBM · Granite 4.1 | 8.38B | 128K | GQA | 15.6 GB | 8.2 GB | 4.7 GB | 2026-04 |
| Granite 4.1 30BIBM · Granite 4.1 | 28.9B | 128K | GQA | 53.8 GB | 27.3 GB | 14.8 GB | 2026-04 |
| Gemma 4 31BGoogle · Gemma 4 | 31.3B | 256K | GQA | 58.3 GB | 30.4 GB | 17.4 GB | 2026-03 |
| Gemma 4 26B A4BGoogle · Gemma 4 | 25.8B · 3.82B act. | 256K | GQA | 48.1 GB | 24.7 GB | 13.8 GB | 2026-03 |
| Gemma 4 E4BGoogle · Gemma 4 | 8.00B | 128K | GQA | 14.9 GB | 8.1 GB | 4.9 GB | 2026-03 |
| Gemma 4 E2BGoogle · Gemma 4 | 5.12B | 128K | MQA | 9.5 GB | 5.1 GB | 3.1 GB | 2026-03 |
| Mistral Small 4 119BMistral AI · Mistral Small | 119B · 6.50B act. | 256K | MLA | 222 GB | 113 GB | 62.0 GB | 2026-03 |
| Nemotron 3 Nano 4BNVIDIA · Nemotron 3 | 3.97B | 256K | hybrid | 7.4 GB | 5.2 GB | 4.2 GB | 2026-03 |
| Nemotron 3 Super 120B-A12BNVIDIA · Nemotron 3 | 124B · 12.0B act. | 1,024K | hybrid | 230 GB | 117 GB | 64.1 GB | 2026-03 |
| Qwen3-Coder-Next 80B-A3BAlibaba Qwen · Qwen3-Coder | 79.7B · 3.87B act. | 256K | hybrid | 148 GB | 75.4 GB | 41.1 GB | 2026-02 |
| Qwen3.5 397B-A17BAlibaba Qwen · Qwen3.5 | 403B · 17.4B act. | 256K | hybrid | 751 GB | 379 GB | 205 GB | 2026-02 |
| Qwen3.5 122B-A10BAlibaba Qwen · Qwen3.5 | 125B · 9.77B act. | 256K | hybrid | 233 GB | 119 GB | 66.1 GB | 2026-02 |
| Qwen3.5 35B-A3BAlibaba Qwen · Qwen3.5 | 36.0B · 3.45B act. | 256K | hybrid | 67.0 GB | 35.4 GB | 20.6 GB | 2026-02 |
| Qwen3.5 27BAlibaba Qwen · Qwen3.5 | 27.8B | 256K | hybrid | 51.7 GB | 30.6 GB | 20.7 GB | 2026-02 |
| Qwen3.5 9BAlibaba Qwen · Qwen3.5 | 9.65B | 256K | hybrid | 18.0 GB | 12.8 GB | 10.3 GB | 2026-02 |
| Qwen3.5 4BAlibaba Qwen · Qwen3.5 | 4.66B | 256K | hybrid | 8.7 GB | 4.9 GB | 3.2 GB | 2026-02 |
| Qwen3.5 2BAlibaba Qwen · Qwen3.5 | 2.27B | 256K | hybrid | 4.2 GB | 2.6 GB | 1.8 GB | 2026-02 |
| Qwen3.5 0.8BAlibaba Qwen · Qwen3.5 | 0.87B | 256K | hybrid | 1.6 GB | 1.0 GB | 0.8 GB | 2026-02 |
| GLM-5 / GLM-5.1Zhipu AI (Z.ai) · GLM-5 | 754B · 40.0B act. | 198K | MLA | 1,404 GB | 706 GB | 378 GB | 2026-02 |
| Kimi K2.5Moonshot AI · Kimi K2 | 1,027B · 32.9B act. | 256K | MLA | 1,913 GB | 961 GB | 514 GB | 2026-01 |
| GLM-4.7-Flash 30B-A3BZhipu AI (Z.ai) · GLM-4.5 | 31.2B · 3.90B act. | 198K | MLA | 58.2 GB | 30.3 GB | 17.2 GB | 2026-01 |
| DeepSeek V3.2DeepSeek · DeepSeek V3 | 685B · 37.6B act. | 160K | MLA | 1,277 GB | 642 GB | 344 GB | 2025-12 |
| Devstral Small 2 24BMistral AI · Devstral | 24.0B | 384K | GQA | 44.7 GB | 24.9 GB | 15.6 GB | 2025-12 |
| Ministral 3 3BMistral AI · Ministral 3 | 3.85B | 256K | GQA | 7.2 GB | 4.0 GB | 2.5 GB | 2025-12 |
| Ministral 3 8BMistral AI · Ministral 3 | 8.92B | 256K | GQA | 16.6 GB | 10.3 GB | 7.4 GB | 2025-12 |
| Ministral 3 14BMistral AI · Ministral 3 | 13.9B | 256K | GQA | 26.0 GB | 15.5 GB | 10.6 GB | 2025-12 |
| Mistral Large 3 (675B)Mistral AI · Mistral Large | 675B · 40.0B act. | 256K | MLA | 1,257 GB | 632 GB | 339 GB | 2025-12 |
| Devstral 2 123BMistral AI · Devstral | 125B | 256K | GQA | 233 GB | 122 GB | 70.7 GB | 2025-12 |
| GLM-4.7Zhipu AI (Z.ai) · GLM-4.5 | 358B · 33.6B act. | 198K | GQA | 667 GB | 337 GB | 182 GB | 2025-12 |
| Nemotron 3 Nano 30B-A3BNVIDIA · Nemotron 3 | 31.6B · 3.50B act. | 1,024K | hybrid | 58.8 GB | 30.7 GB | 17.6 GB | 2025-12 |
| Olmo 3.1 32BAllenAI · Olmo 3 | 32.2B | 64K | GQA | 60.0 GB | 31.9 GB | 18.8 GB | 2025-12 |
| Kimi K2 ThinkingMoonshot AI · Kimi K2 | 1,026B · 32.9B act. | 256K | MLA | 1,912 GB | 960 GB | 514 GB | 2025-11 |
| Olmo 3 7BAllenAI · Olmo 3 | 7.30B | 64K | MHA | 13.6 GB | 8.3 GB | 5.9 GB | 2025-11 |
| Kimi Linear 48B-A3BMoonshot AI · Kimi Linear | 49.1B · 3.48B act. | 1,024K | hybrid | 91.5 GB | 47.2 GB | 26.4 GB | 2025-10 |
| MiniMax-M2 / M2.1 / M2.5MiniMax · MiniMax | 229B · 11.0B act. | 192K | GQA | 426 GB | 215 GB | 117 GB | 2025-10 |
| Granite 4.0 H-Small 32B-A9BIBM · Granite 4.0 | 32.2B · 8.80B act. | 128K | hybrid | 60.0 GB | 30.4 GB | 16.5 GB | 2025-10 |
| Granite 4.0 H-Tiny 7B-A1BIBM · Granite 4.0 | 6.94B · 1.47B act. | 128K | hybrid | 12.9 GB | 6.6 GB | 3.6 GB | 2025-10 |
| Granite 4.0 H-Micro 3BIBM · Granite 4.0 | 3.19B | 128K | hybrid | 5.9 GB | 3.2 GB | 1.9 GB | 2025-10 |
| Qwen3-Next 80B-A3BAlibaba Qwen · Qwen3-Next | 81.3B · 3.87B act. | 256K | hybrid | 151 GB | 76.9 GB | 41.9 GB | 2025-09 |
| Magistral Small 1.2 (2509)Mistral AI · Mistral Small | 24.0B | 128K | GQA | 44.7 GB | 24.9 GB | 15.6 GB | 2025-09 |
| Kimi K2 Instruct (0905)Moonshot AI · Kimi K2 | 1,026B · 32.9B act. | 256K | MLA | 1,912 GB | 960 GB | 514 GB | 2025-09 |
| GLM-4.6Zhipu AI (Z.ai) · GLM-4.5 | 357B · 33.6B act. | 198K | GQA | 665 GB | 335 GB | 181 GB | 2025-09 |
| Gemma 3 270MGoogle · Gemma 3 | 0.27B | 32K | MQA | 0.5 GB | 0.4 GB | 0.4 GB | 2025-08 |
| DeepSeek V3.1DeepSeek · DeepSeek V3 | 685B · 37.6B act. | 160K | MLA | 1,275 GB | 641 GB | 344 GB | 2025-08 |
| gpt-oss-120bOpenAI · gpt-oss | 117B · 5.13B act. | 128K | GQA | 218 GB | 111 GB | 61.0 GB | 2025-08 |
| gpt-oss-20bOpenAI · gpt-oss | 20.9B · 3.61B act. | 128K | GQA | 38.9 GB | 21.6 GB | 13.5 GB | 2025-08 |
| Nemotron Nano 9B v2NVIDIA · Nemotron | 8.89B | 128K | hybrid | 16.6 GB | 10.5 GB | 7.6 GB | 2025-08 |
| Nemotron Nano 12B v2NVIDIA · Nemotron | 12.3B | 128K | hybrid | 22.9 GB | 14.0 GB | 9.8 GB | 2025-08 |
| Qwen3-Coder 30B-A3BAlibaba Qwen · Qwen3-Coder | 30.5B · 3.35B act. | 256K | GQA | 56.9 GB | 29.6 GB | 16.8 GB | 2025-07 |
| Qwen3-Coder 480B-A35BAlibaba Qwen · Qwen3-Coder | 480B · 35.5B act. | 256K | GQA | 894 GB | 451 GB | 243 GB | 2025-07 |
| GLM-4.5Zhipu AI (Z.ai) · GLM-4.5 | 358B · 33.6B act. | 128K | GQA | 667 GB | 337 GB | 182 GB | 2025-07 |
| GLM-4.5-AirZhipu AI (Z.ai) · GLM-4.5 | 110B · 13.4B act. | 128K | GQA | 206 GB | 105 GB | 58.1 GB | 2025-07 |
| Llama-3.3-Nemotron-Super-49B v1.5NVIDIA · Nemotron | 49.9B | 128K | GQA | 92.9 GB | 50.4 GB | 30.4 GB | 2025-07 |
| SmolLM3 3BHugging Face · SmolLM | 3.08B | 128K | GQA | 5.7 GB | 3.1 GB | 1.9 GB | 2025-07 |
| Gemma 3n E4BGoogle · Gemma 3n | 7.85B | 32K | GQA | 14.6 GB | 7.8 GB | 4.6 GB | 2025-06 |
| Gemma 3n E2BGoogle · Gemma 3n | 5.44B | 32K | GQA | 10.1 GB | 5.6 GB | 3.4 GB | 2025-06 |
| Mistral Small 3.2 24BMistral AI · Mistral Small | 24.0B | 128K | GQA | 44.7 GB | 24.9 GB | 15.6 GB | 2025-06 |
| MiniMax-M1 (80k)MiniMax · MiniMax | 456B · 45.9B act. | 977K | hybrid | 850 GB | 429 GB | 232 GB | 2025-06 |
| DeepSeek R1-0528DeepSeek · DeepSeek R1 | 685B · 37.6B act. | 160K | MLA | 1,275 GB | 641 GB | 344 GB | 2025-05 |
| Llama 4 Scout (17B-16E)Meta · Llama 4 | 109B · 17.2B act. | 10,240K | GQA | 202 GB | 105 GB | 59.4 GB | 2025-04 |
| Llama 4 Maverick (17B-128E)Meta · Llama 4 | 402B · 17.2B act. | 1,024K | GQA | 748 GB | 378 GB | 204 GB | 2025-04 |
| Qwen3 0.6BAlibaba Qwen · Qwen3 | 0.60B | 128K | GQA | 1.1 GB | 0.7 GB | 0.5 GB | 2025-04 |
| Qwen3 1.7BAlibaba Qwen · Qwen3 | 1.72B | 128K | GQA | 3.2 GB | 1.9 GB | 1.3 GB | 2025-04 |
| Qwen3 4BAlibaba Qwen · Qwen3 | 4.02B | 128K | GQA | 7.5 GB | 4.1 GB | 2.5 GB | 2025-04 |
| Qwen3 8BAlibaba Qwen · Qwen3 | 8.19B | 128K | GQA | 15.3 GB | 9.9 GB | 7.5 GB | 2025-04 |
| Qwen3 14BAlibaba Qwen · Qwen3 | 14.8B | 128K | GQA | 27.5 GB | 16.7 GB | 11.6 GB | 2025-04 |
| Qwen3 32BAlibaba Qwen · Qwen3 | 32.8B | 128K | GQA | 61.0 GB | 33.4 GB | 20.5 GB | 2025-04 |
| Qwen3 30B-A3BAlibaba Qwen · Qwen3 | 30.5B · 3.35B act. | 128K | GQA | 56.9 GB | 29.6 GB | 16.8 GB | 2025-04 |
| Qwen3 235B-A22BAlibaba Qwen · Qwen3 | 235B · 22.2B act. | 256K | GQA | 438 GB | 221 GB | 120 GB | 2025-04 |
| Phi-4-reasoning 14BMicrosoft · Phi-4 | 14.7B | 32K | GQA | 27.3 GB | 15.6 GB | 10.1 GB | 2025-04 |
| Llama-3.1-Nemotron-Ultra-253BNVIDIA · Nemotron | 253B | 128K | GQA | 472 GB | 244 GB | 137 GB | 2025-04 |
| Granite 3.3 8BIBM · Granite 3.3 | 8.17B | 128K | GQA | 15.2 GB | 7.8 GB | 4.3 GB | 2025-04 |
| Gemma 3 27BGoogle · Gemma 3 | 27.4B | 128K | GQA | 51.1 GB | 26.9 GB | 15.5 GB | 2025-03 |
| Gemma 3 12BGoogle · Gemma 3 | 12.2B | 128K | GQA | 22.7 GB | 12.3 GB | 7.4 GB | 2025-03 |
| Gemma 3 4BGoogle · Gemma 3 | 4.30B | 128K | GQA | 8.0 GB | 4.6 GB | 3.0 GB | 2025-03 |
| Gemma 3 1BGoogle · Gemma 3 | 1.00B | 32K | MQA | 1.9 GB | 1.2 GB | 0.9 GB | 2025-03 |
| Command A (03-2025)Cohere · Command | 111B | 256K | GQA | 207 GB | 106 GB | 59.3 GB | 2025-03 |
| OLMo 2 32BAllenAI · OLMo 2 | 32.2B | 4K | GQA | 60.0 GB | 31.9 GB | 18.8 GB | 2025-03 |
| Phi-4-mini 3.8BMicrosoft · Phi-4 | 3.84B | 128K | GQA | 7.1 GB | 4.1 GB | 2.7 GB | 2025-02 |
| DeepSeek R1DeepSeek · DeepSeek R1 | 684B · 37.6B act. | 160K | MLA | 1,275 GB | 641 GB | 344 GB | 2025-01 |
| R1-Distill-Qwen 1.5BDeepSeek · DeepSeek R1 Distill | 1.78B | 128K | GQA | 3.3 GB | 2.5 GB | 2.2 GB | 2025-01 |
| R1-Distill-Qwen 7BDeepSeek · DeepSeek R1 Distill | 7.62B | 128K | GQA | 14.2 GB | 9.1 GB | 6.8 GB | 2025-01 |
| R1-Distill-Qwen 14BDeepSeek · DeepSeek R1 Distill | 14.8B | 128K | GQA | 27.5 GB | 16.7 GB | 11.6 GB | 2025-01 |
| R1-Distill-Qwen 32BDeepSeek · DeepSeek R1 Distill | 32.8B | 128K | GQA | 61.0 GB | 33.4 GB | 20.5 GB | 2025-01 |
| R1-Distill-Llama 8BDeepSeek · DeepSeek R1 Distill | 8.03B | 128K | GQA | 15.0 GB | 9.4 GB | 6.8 GB | 2025-01 |
| R1-Distill-Llama 70BDeepSeek · DeepSeek R1 Distill | 70.6B | 128K | GQA | 131 GB | 69.6 GB | 40.7 GB | 2025-01 |
| InternLM3 8BShanghai AI Lab · InternLM | 8.80B | 32K | GQA | 16.4 GB | 10.2 GB | 7.2 GB | 2025-01 |
| Llama 3.3 70BMeta · Llama 3.3 | 70.6B | 128K | GQA | 131 GB | 69.6 GB | 40.7 GB | 2024-12 |
| DeepSeek V3DeepSeek · DeepSeek V3 | 684B · 37.6B act. | 160K | MLA | 1,275 GB | 641 GB | 344 GB | 2024-12 |
| Phi-4 14BMicrosoft · Phi-4 | 14.7B | 16K | GQA | 27.3 GB | 15.6 GB | 10.1 GB | 2024-12 |
| Falcon 3 7BTII · Falcon 3 | 7.46B | 32K | GQA | 13.9 GB | 8.4 GB | 5.9 GB | 2024-12 |
| Falcon 3 10BTII · Falcon 3 | 10.3B | 32K | GQA | 19.2 GB | 11.1 GB | 7.3 GB | 2024-12 |
| Qwen2.5-Coder 32BAlibaba Qwen · Qwen2.5 | 32.8B | 128K | GQA | 61.0 GB | 33.4 GB | 20.5 GB | 2024-11 |
| Mistral Large 2 (2411)Mistral AI · Mistral Large | 123B | 128K | GQA | 228 GB | 116 GB | 62.9 GB | 2024-11 |
| OLMo 2 7BAllenAI · OLMo 2 | 7.30B | 4K | MHA | 13.6 GB | 8.3 GB | 5.9 GB | 2024-11 |
| OLMo 2 13BAllenAI · OLMo 2 | 13.7B | 4K | MHA | 25.5 GB | 14.7 GB | 9.6 GB | 2024-11 |
| Llama 3.2 1BMeta · Llama 3.2 | 1.24B | 128K | GQA | 2.3 GB | 1.4 GB | 1.0 GB | 2024-09 |
| Llama 3.2 3BMeta · Llama 3.2 | 3.21B | 128K | GQA | 6.0 GB | 3.4 GB | 2.1 GB | 2024-09 |
| Qwen2.5 7BAlibaba Qwen · Qwen2.5 | 7.62B | 128K | GQA | 14.2 GB | 9.1 GB | 6.8 GB | 2024-09 |
| Qwen2.5 14BAlibaba Qwen · Qwen2.5 | 14.8B | 128K | GQA | 27.5 GB | 16.7 GB | 11.6 GB | 2024-09 |
| Qwen2.5 32BAlibaba Qwen · Qwen2.5 | 32.8B | 128K | GQA | 61.0 GB | 33.4 GB | 20.5 GB | 2024-09 |
| Qwen2.5 72BAlibaba Qwen · Qwen2.5 | 72.7B | 128K | GQA | 135 GB | 72.4 GB | 42.8 GB | 2024-09 |
| Command R (08-2024)Cohere · Command | 32.3B | 128K | GQA | 60.2 GB | 32.0 GB | 18.8 GB | 2024-08 |
| Command R+ (08-2024)Cohere · Command | 104B | 128K | GQA | 193 GB | 99.6 GB | 55.7 GB | 2024-08 |
| Gemma 2 2BGoogle · Gemma 2 | 2.61B | 8K | GQA | 4.9 GB | 3.0 GB | 2.1 GB | 2024-07 |
| Llama 3.1 8BMeta · Llama 3.1 | 8.03B | 128K | GQA | 15.0 GB | 9.4 GB | 6.8 GB | 2024-07 |
| Llama 3.1 70BMeta · Llama 3.1 | 70.6B | 128K | GQA | 131 GB | 69.6 GB | 40.7 GB | 2024-07 |
| Llama 3.1 405BMeta · Llama 3.1 | 406B | 128K | GQA | 756 GB | 386 GB | 212 GB | 2024-07 |
| Mistral Nemo 12BMistral AI · Mistral | 12.2B | 128K | GQA | 22.8 GB | 13.9 GB | 9.7 GB | 2024-07 |
| Gemma 2 27BGoogle · Gemma 2 | 27.2B | 8K | GQA | 50.7 GB | 26.5 GB | 15.1 GB | 2024-06 |
| Gemma 2 9BGoogle · Gemma 2 | 9.24B | 8K | GQA | 17.2 GB | 9.5 GB | 5.8 GB | 2024-06 |
| Mistral 7B v0.3Mistral AI · Mistral | 7.25B | 32K | GQA | 13.5 GB | 7.2 GB | 4.3 GB | 2024-05 |
| Codestral 22BMistral AI · Mistral | 22.2B | 32K | GQA | 41.4 GB | 21.5 GB | 12.1 GB | 2024-05 |
| Yi-1.5 34B01.AI · Yi | 34.4B | 4K | GQA | 64.1 GB | 33.7 GB | 19.5 GB | 2024-05 |
| Mixtral 8x22BMistral AI · Mixtral | 141B · 39.2B act. | 64K | GQA | 262 GB | 132 GB | 70.7 GB | 2024-04 |
| Mixtral 8x7BMistral AI · Mixtral | 46.7B · 12.9B act. | 32K | GQA | 87.0 GB | 44.0 GB | 23.8 GB | 2023-12 |
| Olmo Hybrid 7BAllenAI · Olmo 3 | 7.43B | 64K | hybrid | 13.8 GB | 8.4 GB | 5.8 GB |
Weights only, with embeddings kept in 16-bit for quantized formats. KV cache and runtime come on top: open a model for the full picture.