DeepSeek-V3.2 vs Kimi K3
DeepSeek-V3.2 (DeepSeek) and Kimi K3 (Moonshot) compared on benchmarks, pricing, context window, and use-case rankings.
- Kimi K3 leads on overall intelligence (ECI 158 vs 146).
- DeepSeek-V3.2 is cheaper on input tokens — $0.25 vs $2.10 per 1M.
- Kimi K3 has a larger context window (1M vs 164K tokens).
- Kimi K3 ranks higher for coding (97th vs 37th percentile).
| Attribute | DeepSeek-V3.2 | Kimi K3 |
|---|---|---|
| Scores | ||
| Intelligence (ECI) | 146 | 158 |
| Coding | 37 | 97 |
| Math | 31 | 80 |
| Reasoning & Knowledge | 30 | 73 |
| Agentic & Tools | 32 | 80 |
| Specifications | ||
| Developer | DeepSeek | Moonshot |
| Family | DeepSeek | Kimi |
| Released | Dec 1, 2025 | Jul 16, 2026 |
| Parameters | — | 2.8T |
| Availability | Open weights (unrestricted) | Open weights (non-commercial) |
| Context window | 164K | 1M |
| Price — $/M input | $0.25 | $2.10 |
| Price — $/M output | $0.42 | $10.95 |
| Inputs | text | text, image, video |
| Outputs | text | text |
| Benchmarks | ||
| AIME 2024/2025 | 88% | 97% |
| APEX | 7% | 39% |
| ARC-AGI | 57% | 95% |
| ARC-AGI-2 | 4% | 60% |
| FrontierMath | 22% | — |
| FrontierMath Tier 4 | 2% | — |
| GPQA Diamond | 83% | 93% |
| Humanity's Last Exam | 25% | 47% |
| LiveCodeBench | 86% | — |
| MMLU-Pro | 86% | — |
| SciCode | 39% | 60% |
| SimpleBench | 53% | 61% |
| SimpleQA Verified | 28% | 51% |
| Terminal-Bench | 40% | — |
| WebDev Arena | 1325 | 1674 |
| WeirdML | 47% | 83% |
| τ²-bench | 91% | — |
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 DeepSeek-V3.2 better than Kimi K3?
On Epoch AI's Capabilities Index, Kimi K3 scores higher (158) than DeepSeek-V3.2 (146). The right pick depends on your task — compare their coding, math, and reasoning scores in the table above.
Which is cheaper, DeepSeek-V3.2 or Kimi K3?
DeepSeek-V3.2 is cheaper on input tokens at $0.25 per million, versus $2.10 (representative OpenRouter pricing).
Which has a larger context window, DeepSeek-V3.2 or Kimi K3?
Kimi K3 supports up to 1M tokens, compared with 164K for the other.
Which is better for coding, DeepSeek-V3.2 or Kimi K3?
Across coding benchmarks like SWE-bench Verified and Terminal-Bench, Kimi K3 ranks higher — 97th vs 37th 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.