Open models you can run locally
Popular open-weight models with their size, licence and the memory they need at each quantization.
Last updated: 2026-10-13 · JSON
| Model | Developer | Parameters | Q4_K_M @ 8K | Licence |
|---|---|---|---|---|
| DeepSeek-R1-0528-Qwen3-8B | DeepSeek | 8.19B | 6.5 GiB | MIT |
| DeepSeek-R1-Distill-Qwen-32B | DeepSeek | 32.76B | 21.9 GiB | MIT |
| DeepSeek-R1-0528 (671B) | DeepSeek | 671B (37B active) | 398.8 GiB | MIT |
| DeepSeek-V4-Flash | DeepSeek | 284B (13B active) | 168.9 GiB | MIT |
| Devstral Small 2 24B Instruct (2512) | Mistral AI | 24B | 16.0 GiB | Apache-2.0 |
| Gemma 3 12B IT | Google DeepMind | 12B | 8.4 GiB | Gemma Terms of Use |
| Gemma 3 27B IT | Google DeepMind | 27B | 17.5 GiB | Gemma Terms of Use |
| Gemma 4 E4B IT | Google DeepMind | 8B | 5.5 GiB | Apache-2.0 |
| Gemma 4 12B Unified IT | Google DeepMind | 11.95B | 8.0 GiB | Apache-2.0 |
| Gemma 4 26B A4B IT | Google DeepMind | 25.2B (3.8B active) | 15.7 GiB | Apache-2.0 |
| Gemma 4 31B IT | Google DeepMind | 30.7B | 19.8 GiB | Apache-2.0 |
| GLM-4.5-Air | Zhipu AI (Z.ai) | 106B (12B active) | 64.8 GiB | MIT |
| GLM-4.7-Flash (30B-A3B) | Zhipu AI (Z.ai) | 30B (3B active) | 18.7 GiB | MIT |
| gpt-oss-20b | OpenAI | 21B (3.6B active) | 13.1 GiB | Apache-2.0 |
| gpt-oss-120b | OpenAI | 117B (5.1B active) | 70.1 GiB | Apache-2.0 |
| Kimi K2.6 | Moonshot AI | 1000B (32B active) | 593.9 GiB | Modified MIT |
| Llama 3.1 8B Instruct | Meta | 8B | 6.2 GiB | Llama 3.1 Community License |
| Llama 3.2 3B Instruct | Meta | 3.21B | 3.3 GiB | Llama 3.2 Community License |
| Llama 3.3 70B Instruct | Meta | 70B | 44.5 GiB | Llama 3.3 Community License |
| Llama 4 Scout 17B-16E Instruct | Meta | 109B (17B active) | 66.6 GiB | Llama 4 Community License |
| Magistral Small 1.2 (2509) | Mistral AI | 24B | 16.0 GiB | Apache-2.0 |
| Mistral Small 3.2 24B Instruct (2506) | Mistral AI | 24B | 16.0 GiB | Apache-2.0 |
| Mistral Small 4 119B (2603) | Mistral AI | 119B (6.5B active) | 71.2 GiB | Apache-2.0 |
| Phi-4 (14B) | Microsoft | 14B | 10.4 GiB | MIT |
| Qwen3-8B | Alibaba Qwen | 8.2B | 6.5 GiB | Apache-2.0 |
| Qwen3-30B-A3B (2507) | Alibaba Qwen | 30.5B (3.3B active) | 19.3 GiB | Apache-2.0 |
| Qwen3-32B | Alibaba Qwen | 32.8B | 21.9 GiB | Apache-2.0 |
| Qwen3-Coder-30B-A3B-Instruct | Alibaba Qwen | 30.5B (3.3B active) | 19.3 GiB | Apache-2.0 |
| Qwen3-Coder-Next (80B-A3B) | Alibaba Qwen | 80B (3B active) | 48.1 GiB | Apache-2.0 |
| Qwen3.5-9B | Alibaba Qwen | 9B | 6.1 GiB | Apache-2.0 |
| Qwen3.5-27B | Alibaba Qwen | 27B | 17.0 GiB | Apache-2.0 |
| Qwen3.5-35B-A3B | Alibaba Qwen | 35B (3B active) | 21.4 GiB | Apache-2.0 |
| Qwen3.5-122B-A10B | Alibaba Qwen | 122B (10B active) | 73.0 GiB | Apache-2.0 |
| Qwen3.6-27B | Alibaba Qwen | 27B | 17.0 GiB | Apache-2.0 |
| Qwen3.6-35B-A3B | Alibaba Qwen | 35B (3B active) | 21.4 GiB | Apache-2.0 |
| Qwen3.8-27B | Alibaba Qwen | 27B | 17.0 GiB | Apache-2.0 |
| Qwen3.8-Flash-Next | Alibaba Qwen | 180B (6B active) | 107.4 GiB | Qwen Community License 1.0 (qwen-community-1.0) |