Claude Opus 4.6 vs GPT-5.3-Codex
Claude Opus 4.6 
GPT-5.3-Codex Verdict
- Claude Opus 4.6 and GPT-5.3-Codex are level on quality (63.3 vs 62.5).
- Claude Opus 4.6 is stronger in agents & tools, coding, human preference, knowledge, maths, multimodal, reasoning.
- GPT-5.3-Codex is stronger in composite, instruction following, long context.
- GPT-5.3-Codex is 2.1× cheaper ($4.81 vs $10.00 per 1M blended).
- GPT-5.3-Codex streams 3.3× faster (121 vs 36 tokens per second).
| Metric | Claude Opus 4.6 | GPT-5.3-Codex |
|---|---|---|
| BenchLeader Index | 63.3 | 62.5 |
| Agents & tools score | 68.8 | 62.7 |
| Coding score | 62.4 | 54.8 |
| Composite score | 69.4 | 70.1 |
| Human preference score | 64.9 | – |
| Instruction following score | 53.9 | 73.2 |
| Knowledge score | 70.6 | 69.3 |
| Long context score | 65.5 | 68.3 |
| Maths score | 65.7 | – |
| Multimodal score | 63.6 | 63.0 |
| Reasoning score | 62.6 | – |
| Blended price $/M | $10.00 | $4.81 |
| Output speed | 36 tok/s | 121 tok/s |
| Time to first answer | 2.4 s | 58.2 s |
| Context window | 1M | 400k |
| GPQA Diamond | 90.5% | – |
| OTIS Mock AIME | 94.4% | – |
| SWE-bench Verified (Epoch) | 78.7% | – |
| Humanity's Last Exam | 19.0% | – |
| Terminal-Bench | 79.8% | 78.4% |
| SimpleBench | 67.6% | – |
| Cybench | 93.0% | – |
| Remote Labor Index | 4.2% | – |
| WeirdML | 77.9% | 79.3% |
| APEX-Agents | 32.4% | 31.8% |
| FrontierCode | 26.6% | – |
| GSO-Bench | 33.3% | – |
| Epoch Capabilities Index | 155.3 | 156.6 |
| LMArena Text | 1498 | – |
| LMArena Hard Prompts | 1527 | – |
| LMArena Coding | 1546 | – |
| LMArena WebDev | 1537 | 1409 |
| LMArena Vision | 1311 | – |
| AA Intelligence Index | 31.9 | 32.5 |
| IFBench | 53.1% | 75.4% |
| AA-LCR | 78.0% | 83.3% |
| MMMU-Pro | 75.4% | 78.5% |
| AA-Omniscience | 13.7 | 10.9 |
| Terminal-Bench Hard | 48.5% | 53.0% |
| GPQA Diamond (AA) | 89.6% | 91.5% |
| Humanity's Last Exam (AA) | 39.9% | 42.5% |
| τ²-Bench Telecom (AA) | 92.1% | 86.0% |
| HiL-Bench | 38.3% | 4.3% |
| EQ-Bench 4 | 1223 | – |
| Kagi LLM Benchmark | 72.4% | – |
| SWE-bench Verified (bash only) | 75.6% | – |
| SWE-bench Verified (any scaffold) | 75.6% | – |
Data as of 2026-09-10. Best configuration of each model; every score links to its source on the model pages.
Claude Opus 4.6 vs GPT-5.3-Codex: questions
- Is Claude Opus 4.6 better than GPT-5.3-Codex?
- Claude Opus 4.6 and GPT-5.3-Codex are level on quality (63.3 vs 62.5). The BenchLeader Index combines every independent quality benchmark; Claude Opus 4.6 is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is Claude Opus 4.6 better than GPT-5.3-Codex for coding?
- Claude Opus 4.6 scores higher in coding (62 vs 55 on the category index, where 50 is average).
- Is Claude Opus 4.6 better than GPT-5.3-Codex for agentic tasks?
- Claude Opus 4.6 scores higher in agentic tasks (69 vs 63 on the category index, where 50 is average).
- Which is cheaper, Claude Opus 4.6 or GPT-5.3-Codex?
- GPT-5.3-Codex is cheaper: $4.81 against $10.00 per million tokens, blended at three input tokens per output token.
- Which is faster, Claude Opus 4.6 or GPT-5.3-Codex?
- GPT-5.3-Codex streams faster: 121 against 36 output tokens per second.
- Which has the larger context window?
- Claude Opus 4.6 accepts more context: 1M against 400k tokens.