GPT-5.2 vs GPT-5.6 Terra
GPT-5.2 Verdict
- GPT-5.6 Terra (xhigh) leads on quality: 64.4 vs 61.6.
- GPT-5.2 is stronger in human preference, instruction following, knowledge, long context, reasoning.
- GPT-5.6 Terra (xhigh) is stronger in agents & tools, coding, composite, multimodal, maths.
- They cost about the same ($4.50 per 1M blended).
| Metric | GPT-5.2 | GPT-5.6 Terra (xhigh) |
|---|---|---|
| BenchLeader Index | 61.6 | 64.4 |
| Agents & tools score | 61.2 | 69.8 |
| Coding score | 53.7 | 63.7 |
| Composite score | 67.5 | 77.3 |
| Human preference score | 67.9 | 66.7 |
| Instruction following score | 67.4 | 65.3 |
| Knowledge score | 61.1 | 61.0 |
| Long context score | 67.9 | 66.0 |
| Multimodal score | 59.9 | 63.0 |
| Reasoning score | 61.3 | 58.6 |
| Maths score | – | 71.0 |
| Blended price $/M | $4.81 | $4.50 |
| Output speed | 68 tok/s | 79 tok/s |
| Time to first answer | 129.4 s | 33.6 s |
| Context window | 400k | 1M |
| Humanity's Last Exam | 27.8% | – |
| Terminal-Bench | 64.9% | – |
| SimpleBench | 45.8% | 48.9% |
| SciCode | – | 51.6% |
| Remote Labor Index | 2.1% | – |
| APEX-Agents | 23.0% | – |
| ProofBench | – | 74.0% |
| Epoch Capabilities Index | 153.5 | – |
| LMArena Text | 1476 | 1466 |
| LMArena Hard Prompts | 1497 | 1491 |
| LMArena Coding | 1515 | 1519 |
| LMArena WebDev | 1417 | 1521 |
| LMArena Vision | 1268 | 1269 |
| LMArena Agent | – | 1.5 |
| AA Intelligence Index | 30.4 | 38.2 |
| IFBench | 75.4% | 66.3% |
| AA-LCR | 82.7% | 79.0% |
| MMMU-Pro | – | 79.5% |
| AA-Omniscience | -0.9 | -3.0 |
| Terminal-Bench Hard | 47.0% | 62.9% |
| GPQA Diamond (AA) | 90.3% | 90.8% |
| Humanity's Last Exam (AA) | 37.7% | 41.9% |
| SciCode (AA) | – | 52.3% |
| τ²-Bench Telecom (AA) | 84.8% | 80.4% |
| LiveCodeBench | – | 85.9% |
| MMLU-Pro | – | 86.7% |
| IOI | – | 65.3% |
| CorpFin | – | 65.3% |
| TaxEval | – | 76.2% |
| GPQA Diamond (Vals) | – | 90.9% |
| SWE-Bench Pro | 29.9% | – |
| VISTA | 46.6% | – |
| MultiNRC | 42.2% | – |
| TutorBench | 53.5% | – |
| Kagi LLM Benchmark | 73.3% | – |
| SWE-bench Verified (bash only) | 69.0% | – |
| ARC-AGI-1 | 94.5% | 94.0% |
| ARC-AGI-2 | 72.9% | 74.2% |
| ARC-AGI-3 | – | 0.7% |
| 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 GPT-5.6 Terra: questions
- Is GPT-5.2 better than GPT-5.6 Terra?
- GPT-5.6 Terra (xhigh) leads on quality: 64.4 vs 61.6. The BenchLeader Index combines every independent quality benchmark; GPT-5.6 Terra (xhigh) is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is GPT-5.2 better than GPT-5.6 Terra for coding?
- GPT-5.6 Terra scores higher in coding (64 vs 54 on the category index, where 50 is average).
- Is GPT-5.2 better than GPT-5.6 Terra for agentic tasks?
- GPT-5.6 Terra scores higher in agentic tasks (70 vs 61 on the category index, where 50 is average).
- Which is cheaper, GPT-5.2 or GPT-5.6 Terra?
- GPT-5.6 Terra is cheaper: $4.50 against $4.81 per million tokens, blended at three input tokens per output token.
- Which is faster, GPT-5.2 or GPT-5.6 Terra?
- GPT-5.6 Terra streams faster: 79 against 68 output tokens per second.
- Which has the larger context window?
- GPT-5.6 Terra accepts more context: 1M against 400k tokens.