Claude Opus 4.5 vs GPT-5.2-Codex
Claude Opus 4.5 
GPT-5.2-Codex Verdict
- Claude Opus 4.5 (thinking) and GPT-5.2-Codex are level on quality (60.7 vs 60.8).
- Claude Opus 4.5 (thinking) is stronger in agents & tools, coding, composite, maths, reasoning.
- GPT-5.2-Codex is stronger in instruction following, knowledge, long context, multimodal.
- GPT-5.2-Codex is 2.1× cheaper ($4.81 vs $10.00 per 1M blended).
- Claude Opus 4.5 (thinking) streams 1.8× faster (45 vs 25 tokens per second).
| Metric | Claude Opus 4.5 (thinking) | GPT-5.2-Codex |
|---|---|---|
| BenchLeader Index | 60.7 | 60.8 |
| Agents & tools score | 74.6 | 63.5 |
| Coding score | 56.4 | 55.7 |
| Composite score | 65.8 | 56.8 |
| Instruction following score | 54.8 | 75.1 |
| Knowledge score | 61.0 | 63.1 |
| Long context score | 65.2 | 67.8 |
| Maths score | 64.0 | – |
| Multimodal score | 58.0 | 60.7 |
| Reasoning score | 58.6 | – |
| Blended price $/M | $10.00 | $4.81 |
| Output speed | 45 tok/s | 25 tok/s |
| Time to first answer | 17.3 s | 6.7 s |
| Context window | 200k | 400k |
| Terminal-Bench | – | 66.5% |
| APEX-Agents | – | 27.6% |
| LMArena WebDev | – | 1339 |
| LiveBench | – | 74.0% |
| LiveBench Reasoning | – | 77.7% |
| LiveBench Coding | – | 83.6% |
| LiveBench Agentic Coding | – | 49.4% |
| LiveBench Mathematics | – | 88.8% |
| LiveBench Data Analysis | – | 78.2% |
| LiveBench Language | – | 73.7% |
| AA Intelligence Index | 29.1 | 28.5 |
| IFBench | 58.0% | 77.6% |
| AA-LCR | 77.3% | 82.3% |
| MMMU-Pro | 74.0% | 76.3% |
| AA-Omniscience | 14 | -2.2 |
| Terminal-Bench Hard | 47.0% | 37.1% |
| GPQA Diamond (AA) | 86.6% | 89.9% |
| Humanity's Last Exam (AA) | 30.1% | 35.7% |
| τ²-Bench Telecom (AA) | 89.5% | 92.1% |
| AIME (Vals) | 95.4% | – |
| LiveCodeBench | 83.7% | 88.0% |
| MMLU-Pro | 87.3% | – |
| IOI | 20.3% | – |
| LegalBench | 84.6% | – |
| CorpFin | 65.1% | – |
| TaxEval | 74.9% | – |
| MedQA | 95.9% | – |
| MGSM | 95.2% | – |
| SWE-bench (Vals) | 76.4% | – |
| GPQA Diamond (Vals) | 85.9% | – |
| SWE-Bench Pro | – | 41.0% |
| MultiChallenge | 59.0% | – |
| PRBench Finance | 46.2% | – |
| PRBench Legal | 44.2% | – |
| VISTA | 46.4% | – |
| MultiNRC | 48.6% | – |
| TutorBench | 51.2% | – |
| Kagi LLM Benchmark | 80.2% | – |
| SWE-bench Verified (bash only) | – | 72.8% |
| SWE-bench Verified (any scaffold) | – | 72.8% |
| ARC-AGI-1 | 80.0% | – |
| ARC-AGI-2 | 37.6% | – |
Data as of 2026-09-10. Best configuration of each model; every score links to its source on the model pages.
Claude Opus 4.5 vs GPT-5.2-Codex: questions
- Is Claude Opus 4.5 better than GPT-5.2-Codex?
- Claude Opus 4.5 (thinking) and GPT-5.2-Codex are level on quality (60.7 vs 60.8). The BenchLeader Index combines every independent quality benchmark; GPT-5.2-Codex is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is Claude Opus 4.5 better than GPT-5.2-Codex for coding?
- Claude Opus 4.5 scores higher in coding (56 vs 56 on the category index, where 50 is average).
- Is Claude Opus 4.5 better than GPT-5.2-Codex for agentic tasks?
- Claude Opus 4.5 scores higher in agentic tasks (75 vs 64 on the category index, where 50 is average).
- Which is cheaper, Claude Opus 4.5 or GPT-5.2-Codex?
- GPT-5.2-Codex is cheaper: $4.81 against $10.00 per million tokens, blended at three input tokens per output token.
- Which is faster, Claude Opus 4.5 or GPT-5.2-Codex?
- Claude Opus 4.5 streams faster: 45 against 25 output tokens per second.
- Which has the larger context window?
- GPT-5.2-Codex accepts more context: 400k against 200k tokens.