Kimi-K3.1-Pro
Side-by-side specs, memory needs, licence, runtimes and benchmarks. Add up to 4 models.
--vram
--gpus
--quant
--add
On your GPU
- Kimi-K3.1-Pro
| Property | |
|---|---|
| Overview | |
| Task | Vision |
| Parameters | 2.78T |
| Active parameters | dense |
| Context length | 1M |
| Input → output | |
| Released | 2 Nov 2026 |
| Licence | kimi-k3.1-pro |
| Gated download | no |
| Memory and speed | |
| Fits your hardwarepick a GPU above | ≈1701 GB |
| VRAM at Q4 | ≈1701 GB |
| VRAM at Q8 | ≈2941 GB |
| VRAM at FP16 | ≈5490 GB |
| Est. generation speedpick a GPU above | — |
| Popularity | |
| Downloads / 30 days | 64.6K |
| Likes | 2.4K |
| Benchmarks | |
| GPQA Diamond | 94.2%self-reported |
| Humanity's Last Exam | 51.1%self-reported |
| MMMU-Pro | 84.1%self-reported |
| Runs with | |
| Transformers | |
| Ollama tag | — |
| Architecture | |
| Architecture | KimiK31ForConditionalGeneration |
| Layers | 93 |
| Attention / KV heads | 96 / 96 |
| Vocabulary | 163 840 |
| Weights dtype | bfloat16 |
| GGUF available | yes |
Keep comparing
Memory figures are estimates for an 8K context at the chosen quantisation (Q4_K_M for LLMs by default). Speeds are rough, memory-bound estimates. Self-reported benchmark scores come from the model authors.