Kimi K2.5 vs Kimi K3
Kimi K2.5 (Moonshot) and Kimi K3 (Moonshot) compared on benchmarks, pricing, context window, and use-case rankings.
- Kimi K3 leads on overall intelligence (ECI 157 vs 148).
- Kimi K2.5 is cheaper on input tokens — $0.38 vs $3.00 per 1M.
- Kimi K3 has a larger context window (1M vs 262K tokens).
- Kimi K3 ranks higher for coding (99th vs 50th percentile).
| Attribute | Kimi K2.5 | Kimi K3 |
|---|---|---|
| Scores | ||
| Intelligence (ECI) | 148 | 157 |
| Coding | 50 | 99 |
| Math | 53 | 85 |
| Reasoning & Knowledge | 37 | 80 |
| Agentic & Tools | 54 | 86 |
| Specifications | ||
| Developer | Moonshot | Moonshot |
| Family | Kimi | Kimi |
| Released | Feb 2, 2026 | Jul 16, 2026 |
| Parameters | 1T | 2.8T |
| Availability | Open weights (unrestricted) | Open weights (non-commercial) |
| Context window | 262K | 1M |
| Price — $/M input | $0.38 | $3.00 |
| Price — $/M output | $2.02 | $14.00 |
| Inputs | text, image | text, image |
| Outputs | text | text |
| Benchmarks | ||
| AIME 2024/2025 | 92% | 97% |
| APEX | 14% | 39% |
| ARC-AGI | 65% | 95% |
| ARC-AGI-2 | 12% | 60% |
| FrontierMath | 28% | — |
| FrontierMath Tier 4 | 4% | — |
| GPQA Diamond | 88% | 93% |
| Humanity's Last Exam | 24% | 44% |
| SciCode | 49% | 59% |
| SimpleBench | 47% | — |
| SimpleQA Verified | 34% | 43% |
| SWE-bench Verified | 74% | — |
| Terminal-Bench | 43% | — |
| WebDev Arena | 1431 | 1682 |
| WeirdML | 46% | 83% |
| τ²-bench | 96% | — |
Use-case scores are 0–100 percentile composites across each area’s benchmarks, ranked against every model from the past year. Highlighted cells lead each row. Open a model for the full picture.
Frequently asked questions
Is Kimi K2.5 better than Kimi K3?
On Epoch AI's Capabilities Index, Kimi K3 scores higher (157) than Kimi K2.5 (148). The right pick depends on your task — compare their coding, math, and reasoning scores in the table above.
Which is cheaper, Kimi K2.5 or Kimi K3?
Kimi K2.5 is cheaper on input tokens at $0.38 per million, versus $3.00 (representative OpenRouter pricing).
Which has a larger context window, Kimi K2.5 or Kimi K3?
Kimi K3 supports up to 1M tokens, compared with 262K for the other.
Which is better for coding, Kimi K2.5 or Kimi K3?
Across coding benchmarks like SWE-bench Verified and Terminal-Bench, Kimi K3 ranks higher — 99th vs 50th percentile among the models tracked on Model Beat.
Want a different match-up? Open the compare tool to add or swap models.
More comparisons
Benchmarks & model data from Epoch AI (CC BY); pricing & specs from OpenRouter. ECI = Epoch Capabilities Index.