GPT-5 vs GPT-5.2
GPT-5 Verdict
- GPT-5 and GPT-5.2 are level on quality (60.8 vs 61.6).
- GPT-5 is stronger in coding, knowledge, maths, reasoning.
- GPT-5.2 is stronger in agents & tools, composite, human preference, instruction following, long context, multimodal.
- GPT-5 is 1.4× cheaper ($3.44 vs $4.81 per 1M blended).
| Metric | GPT-5 | GPT-5.2 |
|---|---|---|
| BenchLeader Index | 60.8 | 61.6 |
| Agents & tools score | 51.0 | 61.2 |
| Coding score | 60.9 | 53.7 |
| Composite score | 58.1 | 67.5 |
| Human preference score | 61.7 | 67.8 |
| Instruction following score | 65.0 | 67.4 |
| Knowledge score | 62.2 | 61.1 |
| Long context score | 65.6 | 67.9 |
| Maths score | 72.8 | – |
| Multimodal score | 59.6 | 59.9 |
| Reasoning score | 64.2 | 61.3 |
| Blended price $/M | $3.44 | $4.81 |
| Output speed | 85 tok/s | 68 tok/s |
| Time to first answer | 60.7 s | 129.4 s |
| Context window | 400k | 400k |
| Humanity's Last Exam | – | 27.8% |
| Terminal-Bench | 49.6% | 64.9% |
| SimpleBench | – | 45.8% |
| SciCode | 42.9% | – |
| Remote Labor Index | 1.7% | 2.1% |
| WeirdML | 39.8% | – |
| APEX-Agents | 18.3% | 23.0% |
| Epoch Capabilities Index | 150 | 153.5 |
| LMArena Text | 1427 | 1476 |
| LMArena Hard Prompts | 1449 | 1497 |
| LMArena Coding | 1462 | 1515 |
| LMArena WebDev | – | 1417 |
| LMArena Vision | 1232 | 1268 |
| AA Intelligence Index | 23.0 | 30.4 |
| IFBench | 73.1% | 75.4% |
| AA-LCR | 78.2% | 82.7% |
| MMMU-Pro | 74.2% | – |
| AA-Omniscience | -8.7 | -0.9 |
| Terminal-Bench Hard | 32.6% | 47.0% |
| GPQA Diamond (AA) | 85.3% | 90.3% |
| Humanity's Last Exam (AA) | 28.5% | 37.7% |
| τ²-Bench Telecom (AA) | 84.8% | 84.8% |
| SWE-Bench Pro | – | 29.9% |
| PRBench Finance | 51.3% | – |
| PRBench Legal | 49.0% | – |
| VISTA | 49.7% | 46.6% |
| MultiNRC | 52.1% | 42.2% |
| TutorBench | 55.3% | 53.5% |
| Kagi LLM Benchmark | 72.7% | 73.3% |
| IFEval (HELM) | 87.5% | – |
| Omni-MATH (HELM) | 64.7% | – |
| WildBench (HELM) | 85.7% | – |
| MMLU-Pro (HELM) | 86.3% | – |
| GPQA Diamond (HELM) | 79.1% | – |
| HELM Capabilities mean | 80.7% | – |
| Aider Polyglot | 88.0% | – |
| SWE-bench Verified (bash only) | – | 69.0% |
| SWE-bench Verified (any scaffold) | 75.6% | – |
| ARC-AGI-1 | – | 94.5% |
| ARC-AGI-2 | – | 72.9% |
| BFCL Overall | – | 55.9% |
Data as of 2026-09-12. Best configuration of each model; every score links to its source on the model pages.
GPT-5 vs GPT-5.2: questions
- Is GPT-5 better than GPT-5.2?
- GPT-5 and GPT-5.2 are level on quality (60.8 vs 61.6). The BenchLeader Index combines every independent quality benchmark; GPT-5.2 is ahead overall as of 2026-09-12, but check the category scores for your use.
- Is GPT-5 better than GPT-5.2 for coding?
- GPT-5 scores higher in coding (61 vs 54 on the category index, where 50 is average).
- Is GPT-5 better than GPT-5.2 for agentic tasks?
- GPT-5.2 scores higher in agentic tasks (61 vs 51 on the category index, where 50 is average).
- Which is cheaper, GPT-5 or GPT-5.2?
- GPT-5 is cheaper: $3.44 against $4.81 per million tokens, blended at three input tokens per output token.
- Which is faster, GPT-5 or GPT-5.2?
- GPT-5 streams faster: 85 against 68 output tokens per second.
- Which has the larger context window?
- Both accept 400k tokens of context.