GPT-5.2 vs Kimi K2.6
GPT-5.2 
Kimi K2.6 Verdict
- GPT-5.2 and Kimi K2.6 are level on quality (61.6 vs 60.7).
- GPT-5.2 is stronger in agents & tools, human preference, knowledge, long context.
- Kimi K2.6 is stronger in coding, composite, instruction following, multimodal, reasoning, maths.
- Kimi K2.6 is 2.8× cheaper ($1.71 vs $4.81 per 1M blended).
- GPT-5.2 streams 1.5× faster (68 vs 44 tokens per second).
| Metric | GPT-5.2 | Kimi K2.6 |
|---|---|---|
| BenchLeader Index | 61.6 | 60.7 |
| Agents & tools score | 61.2 | 45.1 |
| Coding score | 53.7 | 60.4 |
| Composite score | 67.5 | 68.6 |
| Human preference score | 67.9 | 60.8 |
| Instruction following score | 67.4 | 73.7 |
| Knowledge score | 61.1 | 58.6 |
| Long context score | 67.9 | 67.1 |
| Multimodal score | 59.9 | 63.7 |
| Reasoning score | 61.3 | 66.0 |
| Maths score | – | 53.8 |
| Blended price $/M | $4.81 | $1.71 |
| Output speed | 68 tok/s | 44 tok/s |
| Time to first answer | 129.4 s | 103.2 s |
| Context window | 400k | 262k |
| GPQA Diamond | – | 90.8% |
| FrontierMath Tiers 1–3 | – | 57.2% |
| FrontierMath Tier 4 | – | 25.6% |
| OTIS Mock AIME | – | 96.1% |
| SWE-bench Verified (Epoch) | – | 76.7% |
| SimpleQA Verified | – | 34.9% |
| Humanity's Last Exam | 27.8% | – |
| Terminal-Bench | 64.9% | – |
| SimpleBench | 45.8% | – |
| OSWorld-Verified 2.0 | – | 4.6% |
| SciCode | – | 53.5% |
| Remote Labor Index | 2.1% | – |
| WeirdML | – | 55.9% |
| APEX-Agents | 23.0% | 18.9% |
| ProofBench | – | 16.0% |
| Epoch Capabilities Index | 153.5 | 151.0 |
| LMArena Text | 1476 | 1461 |
| LMArena Hard Prompts | 1497 | 1485 |
| LMArena Coding | 1515 | 1514 |
| LMArena WebDev | 1417 | 1509 |
| LMArena Vision | 1268 | 1281 |
| AA Intelligence Index | 30.4 | 31.3 |
| IFBench | 75.4% | 76.0% |
| AA-LCR | 82.7% | 81.0% |
| MMMU-Pro | – | 79.4% |
| AA-Omniscience | -0.9 | 5.3 |
| Terminal-Bench Hard | 47.0% | 43.9% |
| GPQA Diamond (AA) | 90.3% | 91.1% |
| Humanity's Last Exam (AA) | 37.7% | 37.5% |
| SciCode (AA) | – | 51.5% |
| τ²-Bench Telecom (AA) | 84.8% | 95.9% |
| LiveCodeBench | – | 86.8% |
| MMLU-Pro | – | 87.6% |
| LegalBench | – | 84.7% |
| CorpFin | – | 66.7% |
| TaxEval | – | 74.7% |
| Terminal-Bench 2.1 (Vals) | – | 53.6% |
| SWE-bench (Vals) | – | 76.2% |
| GPQA Diamond (Vals) | – | 89.1% |
| Vals Index | – | 43.5 |
| SWE-Bench Pro | 29.9% | – |
| VISTA | 46.6% | – |
| MultiNRC | 42.2% | – |
| HiL-Bench | – | 18.7% |
| TutorBench | 53.5% | – |
| EQ-Bench 4 | – | 1202 |
| Kagi LLM Benchmark | 73.3% | – |
| SWE-bench Verified (bash only) | 69.0% | – |
| ARC-AGI-1 | 94.5% | – |
| ARC-AGI-2 | 72.9% | – |
| BFCL Overall | 55.9% | – |
Data as of 2026-09-10. Best configuration of each model; every score links to its source on the model pages.
GPT-5.2 vs Kimi K2.6: questions
- Is GPT-5.2 better than Kimi K2.6?
- GPT-5.2 and Kimi K2.6 are level on quality (61.6 vs 60.7). The BenchLeader Index combines every independent quality benchmark; GPT-5.2 is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is GPT-5.2 better than Kimi K2.6 for coding?
- Kimi K2.6 scores higher in coding (60 vs 54 on the category index, where 50 is average).
- Is GPT-5.2 better than Kimi K2.6 for agentic tasks?
- GPT-5.2 scores higher in agentic tasks (61 vs 45 on the category index, where 50 is average).
- Which is cheaper, GPT-5.2 or Kimi K2.6?
- Kimi K2.6 is cheaper: $1.71 against $4.81 per million tokens, blended at three input tokens per output token.
- Which is faster, GPT-5.2 or Kimi K2.6?
- GPT-5.2 streams faster: 68 against 44 output tokens per second.
- Which has the larger context window?
- GPT-5.2 accepts more context: 400k against 262k tokens.