GPT-5.2-Codex vs Kimi K3
GPT-5.2-Codex 
Kimi K3 Verdict
- Kimi K3 leads on quality: 66.5 vs 60.8.
- GPT-5.2-Codex is stronger in instruction following.
- Kimi K3 is stronger in agents & tools, coding, composite, knowledge, long context, multimodal, human preference, maths.
- GPT-5.2-Codex is 1.2× cheaper ($4.81 vs $6.00 per 1M blended).
- Kimi K3 streams 1.5× faster (38 vs 25 tokens per second).
| Metric | GPT-5.2-Codex | Kimi K3 |
|---|---|---|
| BenchLeader Index | 60.8 | 66.5 |
| Agents & tools score | 63.5 | 67.4 |
| Coding score | 55.7 | 65.0 |
| Composite score | 56.8 | 74.2 |
| Instruction following score | 75.1 | – |
| Knowledge score | 63.1 | 68.2 |
| Long context score | 67.8 | 71.1 |
| Multimodal score | 60.7 | 65.0 |
| Human preference score | – | 71.2 |
| Maths score | – | 78.1 |
| Blended price $/M | $4.81 | $6.00 |
| Output speed | 25 tok/s | 38 tok/s |
| Time to first answer | 6.7 s | 56.3 s |
| Context window | 400k | 1.0M |
| Terminal-Bench | 66.5% | – |
| SciCode | – | 58.7% |
| APEX-Agents | 27.6% | 39.3% |
| FrontierCode | – | 44.2% |
| ProofBench | – | 87.0% |
| Epoch Capabilities Index | – | 157.6 |
| LMArena WebDev | 1339 | – |
| LiveBench | 74.0% | 79.2% |
| LiveBench Reasoning | 77.7% | 90.7% |
| LiveBench Coding | 83.6% | 81.5% |
| LiveBench Agentic Coding | 49.4% | 62.2% |
| LiveBench Mathematics | 88.8% | 84.4% |
| LiveBench Data Analysis | 78.2% | 78.7% |
| LiveBench Language | 73.7% | 85.5% |
| AA Intelligence Index | 28.5 | 43.8 |
| IFBench | 77.6% | – |
| AA-LCR | 82.3% | 88.7% |
| MMMU-Pro | 76.3% | 80.5% |
| AA-Omniscience | -2.2 | 19.7 |
| Terminal-Bench Hard | 37.1% | – |
| GPQA Diamond (AA) | 89.9% | 93.5% |
| Humanity's Last Exam (AA) | 35.7% | 46.9% |
| SciCode (AA) | – | 59.5% |
| τ²-Bench Telecom (AA) | 92.1% | – |
| LiveCodeBench | 88.0% | – |
| LegalBench | – | 86.0% |
| CorpFin | – | 71.6% |
| TaxEval | – | 75.7% |
| Terminal-Bench 2.1 (Vals) | – | 80.9% |
| SWE-bench (Vals) | – | 93.4% |
| GPQA Diamond (Vals) | – | 92.9% |
| Vals Index | – | 57.8 |
| SWE-Bench Pro | 41.0% | – |
| EQ-Bench 4 | – | 1339 |
| SWE-bench Verified (bash only) | 72.8% | – |
| SWE-bench Verified (any scaffold) | 72.8% | – |
Data as of 2026-09-10. Best configuration of each model; every score links to its source on the model pages.
GPT-5.2-Codex vs Kimi K3: questions
- Is GPT-5.2-Codex better than Kimi K3?
- Kimi K3 leads on quality: 66.5 vs 60.8. The BenchLeader Index combines every independent quality benchmark; Kimi K3 is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is GPT-5.2-Codex better than Kimi K3 for coding?
- Kimi K3 scores higher in coding (65 vs 56 on the category index, where 50 is average).
- Is GPT-5.2-Codex better than Kimi K3 for agentic tasks?
- Kimi K3 scores higher in agentic tasks (67 vs 64 on the category index, where 50 is average).
- Which is cheaper, GPT-5.2-Codex or Kimi K3?
- GPT-5.2-Codex is cheaper: $4.81 against $6.00 per million tokens, blended at three input tokens per output token.
- Which is faster, GPT-5.2-Codex or Kimi K3?
- Kimi K3 streams faster: 38 against 25 output tokens per second.
- Which has the larger context window?
- Kimi K3 accepts more context: 1.0M against 400k tokens.