Models
41 open models, smallest first
The machine count is how many of the 36 in the catalog hold the 4-bit file plus 8,192 tokens of context in the memory their GPU can use without raising any limit.
| Model | Maker | Parameters | 4-bit file | Machines it fits |
|---|---|---|---|---|
| Qwen3.5 4B | Alibaba Qwen | 4B | 2.6 GB | 36 |
| Llama 3.1 8B Instruct | Meta | 8B | 4.6 GB | 36 |
| Gemma 4 E4B IT | 8B | 5.0 GB | 36 | |
| Qwen3 8B | Alibaba Qwen | 8.2B | 4.7 GB | 36 |
| Granite 4.2 8B | IBM | 8.8B | 5.2 GB | 36 |
| Qwen3.5 9B | Alibaba Qwen | 9B | 5.3 GB | 36 |
| Gemma 4 12B IT | 11.95B | 6.6 GB | 36 | |
| Ministral 3 14B Reasoning (2512) | Mistral AI | 13.9B | 7.7 GB | 36 |
| Phi-4 | Microsoft | 14.7B | 8.3 GB | 36 |
| gpt-oss-20b | OpenAI | 21B, 3.6B active | 10.8 GB | 36 |
| Devstral Small 2 24B (2512) | Mistral AI | 24B | 13.3 GB | 36 |
| Gemma 4 26B-A4B IT | 25.2B, 3.8B active | 15.9 GB | 36 | |
| Qwen3.8 27B | Alibaba Qwen | 27B | 16.2 GB | 36 |
| Qwen3.6 27B | Alibaba Qwen | 27B | 15.7 GB | 36 |
| Granite 4.2 30B | IBM | 29.3B | 16.8 GB | 36 |
| Nemotron 3.5 Lightning 30B-A3B | NVIDIA | 30B, 3B active | 23.7 GB | 30 |
| Qwen3 Coder 30B-A3B Instruct | Alibaba Qwen | 30.5B, 3.3B active | 17.3 GB | 36 |
| Gemma 4 31B IT | 30.7B | 17.1 GB | 36 | |
| Qwen3.6 35B-A3B | Alibaba Qwen | 35B, 3B active | 20.8 GB | 36 |
| Ornith 1.5 35B-A3B | Ornith (DeepReinforce) | 35B, 3B active | 20.2 GB | 36 |
| Llama 3.3 70B Instruct | Meta | 70B | 39.6 GB | 28 |
| Qwen3 Coder Next | Alibaba Qwen | 80B, 3B active | 45.2 GB | 25 |
| GLM-4.5-Air | Z.ai | 106B, 12B active | 68.0 GB | 22 |
| Llama 4 Scout 17B-16E Instruct | Meta | 109B, 17B active | 60.9 GB | 23 |
| gpt-oss-120b | OpenAI | 117B, 5.1B active | 58.5 GB | 23 |
| Mistral Small 4 119B (2603) | Mistral AI | 119B, 6.5B active | 67.6 GB | 22 |
| Nemotron 3 Super 120B-A12B | NVIDIA | 120B, 12B active | 81.0 GB | 22 |
| Qwen3.5 122B-A10B | Alibaba Qwen | 122B, 10B active | 71.3 GB | 22 |
| Qwen3.8 Flash-Next | Alibaba Qwen | 125B, 6B active | 111.4 GB | 10 |
| Mistral Medium 3.5 128B | Mistral AI | 128B | 69.8 GB | 22 |
| MiniMax M2.7 | MiniMax | 228.7B | 129.3 GB | 6 |
| Hy3 | Tencent | 295B, 21B active | 169.6 GB | 5 |
| GLM-5.3 Flash | Z.ai | 320B, 18B active | 186.0 GB | 4 |
| Qwen3.5 397B-A17B | Alibaba Qwen | 397B, 17B active | 227.3 GB | 2 |
| MiniMax M3 | MiniMax | 428B, 23B active | 243.3 GB | 2 |
| DeepSeek V4.1 Flash | DeepSeek | 552B, 16B active | 483.0 GB | 1 |
| DeepSeek V3.2 | DeepSeek | 671B, 37B active | 377.6 GB | 1 |
| GLM-5.3 | Z.ai | 753.3B | 435.2 GB | 1 |
| Kimi K2.6 | Moonshot AI | 1T, 32B active | 543.6 GB | 1 |
| DeepSeek V4 Pro | DeepSeek | 1.6T, 49B active | 889.0 GB | 0 |
| Kimi K3 | Moonshot AI | 2.8T, 104B active | 1405.1 GB | 0 |