Kimi K2.5
kimi-k2
Earlier Kimi frontier model for long-context agents, coding, and multimodal work
ReasoningTool CallingStructured OutputOpen Weights
Context Window
262K
Max Output
262K
Temperature
No
Open Weights
Yes
Knowledge Cutoff
2025-01
Released
2026-01
Last Updated
2026-01
Modalities
Input:TextImageVideo
→Output:Text
Benchmarks
| Benchmark | Score | Metric | Harness |
|---|---|---|---|
| SWE-Bench Verified | 70.8 | resolved | - |
| SWE-Atlas Codebase QnA | 13.1 | score | Mini-SWE-Agent |
| SWE-Atlas Refactoring | 20.95 | score | Mini-SWE-Agent |
| SWE-Atlas Test Writing | 25.77 | score | Mini-SWE-Agent |
Available Providers (10)
| Provider | Input /1M | Output /1M | Cache Read /1M | Cache Write /1M | Reasoning | Status |
|---|---|---|---|---|---|---|
 Qiniu | — | — | — | — | — | — |
 Vercel AI Gateway | Rp 10.821 | Rp 54.102 | Rp 1.804 | — | toggle | — |
 NovitaAI | Rp 10.821 | Rp 54.102 | Rp 1.804 | — | toggle | — |
 ZenMux | Rp 10.460 | Rp 54.462 | Rp 1.804 | — | toggle | — |
 Merge Gateway | Rp 10.821 | Rp 54.102 | — | — | toggle, budget_tokens | — |
 HPC-AI | Rp 5.411 | Rp 27.051 | Rp 902 | — | toggle | — |
 Kilo Gateway | Rp 8.116 | Rp 39.675 | — | — | toggle | — |
 OpenRouter | Rp 6.763 | Rp 36.519 | — | — | ✓ | — |
 Jiekou.AI | Rp 10.821 | Rp 54.102 | — | — | budget_tokens | — |
 NanoGPT | Rp 5.411 | Rp 34.265 | — | — | — | — |