Model comparison
Kimi K2.6 vs Kimi K2.5
- Input
- $0.95 / 1M tokens
- Output
- $4.00 / 1M tokens
- Context
- 256K
- Released
- 2026-04-20
- Input
- $0.60 / 1M tokens
- Output
- $2.50 / 1M tokens
- Context
- 256K
- Released
- 2026-01-27
Kimi K2.6 scores higher on 5 of the 5 benchmarks both models report; Kimi K2.5 is cheaper on input tokens ($0.6 vs $1 per 1M, 1.6×).
Specs and pricing side by side
| Spec | Kimi K2.6 | Kimi K2.5 |
|---|---|---|
| Maker | Moonshot AI (Kimi) | Moonshot AI (Kimi)same |
| Released | 2026-04-20 | 2026-01-27 |
| Context window | 256K | 256Ksame |
| Max output | 256K | 16K |
| Input price | $0.95 / 1M tokens | $0.60 / 1M tokens |
| Output price | $4.00 / 1M tokens | $2.50 / 1M tokens |
| Parameters | 1T total (MoE), 32B active per token | 1T total, 32B active (MoE) |
| License | Modified MIT (open weights; "Kimi K2" UI attribution required above 100M MAU or $20M monthly revenue) | Modified MIT License |
| Weights | Open weights | Open weightssame |
| Modalities | Text, Vision, Video | Text, Vision, Videosame |
Not published by either maker: knowledge cutoff.
Benchmarks both models report
| Benchmark | Kimi K2.6 | Kimi K2.5 |
|---|---|---|
| SWE-Bench Verified | 80.2% | 76.8% |
| SWE-Bench Pro | 58.6% | 50.7% |
| Humanity's Last Exam | 54% | 30.1% |
| LiveCodeBench v6 | 89.6% | 85% |
| GPQA-Diamond | 90.5% | 87.6% |
Scores are as published by each maker or its cited source; harness, reasoning effort and date can differ between rows, so treat small gaps as ties. Sources are listed on each model page.
Which one should you pick?
Pick Kimi K2.6 if…
- You care about SWE-Bench Verified, SWE-Bench Pro, Humanity's Last Exam (it scores higher on 5 shared benchmarks).
- Your use case is: autonomous, multi-step software engineering and bug-fixing over large repositories.
Pick Kimi K2.5 if…
- Cost matters: input tokens cost 1.6× less ($0.6 vs $1 per 1M).
- Your use case is: agentic software engineering: multi-file bug fixes, refactors, and repo-level tasks.
FAQ
Which is better, Kimi K2.6 or Kimi K2.5?
Kimi K2.6 scores higher on 5 of the 5 benchmarks both models report; Kimi K2.5 is cheaper on input tokens ($0.6 vs $1 per 1M, 1.6×). The better pick depends on whether you value benchmark scores, price per token, context length or openness for your workload.
How much cheaper is Kimi K2.5?
Kimi K2.6: $0.95 / 1M tokens input, $4.00 / 1M tokens output. Kimi K2.5: $0.60 / 1M tokens input, $2.50 / 1M tokens output (list API prices as published by each maker).
What are the context windows of Kimi K2.6 and Kimi K2.5?
Kimi K2.6 offers 256K with up to 256K output; Kimi K2.5 offers 256K with up to 16K output.
Is Kimi K2.6 newer than Kimi K2.5?
Kimi K2.6 was released 2026-04-20; Kimi K2.5 was released 2026-01-27.