Kimi K2.5
Kimi K2.5 is an open-weights language model from Moonshot AI, released January 27, 2026. It has 1T parameters (mixture-of-experts, 33B active). The context window is 256K tokens. On the Epoch Capabilities Index it scores 148.8, rank 75 of 256 rated models.
We introduce Kimi K2.5, an open-source multimodal agentic model designed to advance general agentic intelligence. K2.5 emphasizes the joint optimization of text and vision so that two modalities enhance each other.
Check it against your own PC →
Can you run Kimi K2.5 locally?
Even with 32 GB of VRAM and 32 GB of RAM it is too large to run at a usable speed on a single consumer GPU. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Too large | – | – | – | – |
| 8 GB VRAM | Too large | – | – | – | – |
| 12 GB VRAM | Too large | – | – | – | – |
| 16 GB VRAM | Too large | – | – | – | – |
| 24 GB VRAM | Too large | – | – | – | – |
| 32 GB VRAM | Too large | – | – | – | – |
Ready-made quantizations on Hugging Face: unsloth/Kimi-K2.5-GGUF · bartowski/moonshotai_Kimi-K2.5-GGUF · QuantPasture/Kimi-K2.5-GGUF · DevQuasar/moonshotai.Kimi-K2.5-GGUF · nvidia/Kimi-K2.5-NVFP4 · ISTA-DASLab/Kimi-K2.5-P48NVFP4-MoESQ
Strengths by category
ModelTape's 0–100 fit on Epoch AI benchmark results.
| Category | Score | Based on |
|---|---|---|
| Reasoning | 15.0 | SimpleBench, Humanity's Last Exam |
| Math | 59.7 | OTIS Mock AIME 2024-2025, FrontierMath-2025-02-28-Private, FrontierMath-Tier-4-2025-07-01-Private |
| Coding | 46.1 | Terminal Bench, SWE-bench Verified |
| Knowledge | 69.8 | GPQA diamond, SimpleQA Verified |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| OTIS Mock AIME | 92.2% |
| GPQA Diamond | 86.9% |
| SWE-bench Verified | 73.8% |
| OSWorld | 63.3% |
| FrontierMath-2025-02-28-Private | 48.9% |
| Terminal-Bench | 43.2% |
| SimpleBench | 36.2% |
| SimpleQA Verified | 34.3% |
| Humanity's Last Exam | 24.4% |
| FrontierMath-Tier-4-2025-07-01-Private | 7.0% |
Links
Announcement · Hugging Face: moonshotai/Kimi-K2.5 · Epoch AI benchmarks
Models with a similar score
Mistral Large 4 Preview · Qwen 3.8 27B · Gemini 3 Flash · Grok 4.20 · Claude Opus 4.5 · Kimi K2.6 · GPT-5 Pro · Kimi K2.7 Code
More from Moonshot AI
Kimi K3.1 Pro · Kimi K3 · Kimi K2.7 Code · Kimi K2.6 · Kimi K2 Thinking · Kimi Linear · Kimi K2 (Jul 2025) · Kimi-VL