GPT-5.2-Codex vs Qwen3 7
GPT-5.2-Codex 
Qwen3 7 Verdict
- GPT-5.2-Codex and Qwen3 7 (max) are level on quality (60.8 vs 61.3).
- GPT-5.2-Codex is stronger in agents & tools, composite, long context, multimodal.
- Qwen3 7 (max) is stronger in coding, instruction following, knowledge, human preference, maths, reasoning.
- Qwen3 7 (max) is 1.3× cheaper ($3.75 vs $4.81 per 1M blended).
- Qwen3 7 (max) streams 6.8× faster (169 vs 25 tokens per second).
| Metric | GPT-5.2-Codex | Qwen3 7 (max) |
|---|---|---|
| BenchLeader Index | 60.8 | 61.3 |
| Agents & tools score | 63.5 | 56.0 |
| Coding score | 55.7 | 61.4 |
| Composite score | 56.8 | 56.4 |
| Instruction following score | 75.1 | 77.6 |
| Knowledge score | 63.1 | 63.6 |
| Long context score | 67.8 | 66.0 |
| Multimodal score | 60.7 | – |
| Human preference score | – | 57.2 |
| Maths score | – | 57.4 |
| Reasoning score | – | 66.8 |
| Blended price $/M | $4.81 | $3.75 |
| Output speed | 25 tok/s | 169 tok/s |
| Time to first answer | 6.7 s | 16.5 s |
| Context window | 400k | 1M |
| GPQA Diamond | – | 90.9% |
| FrontierMath Tiers 1–3 | – | 64.6% |
| FrontierMath Tier 4 | – | 34.1% |
| OTIS Mock AIME | – | 95.6% |
| SWE-bench Verified (Epoch) | – | 77.3% |
| SimpleQA Verified | – | 55.8% |
| Terminal-Bench | 66.5% | – |
| SimpleBench | – | 70.4% |
| SciCode | – | 48.8% |
| APEX-Agents | 27.6% | – |
| ProofBench | – | 26.0% |
| Epoch Capabilities Index | – | 153.7 |
| LMArena Text | – | 1474 |
| LMArena Hard Prompts | – | 1495 |
| LMArena Coding | – | 1525 |
| LMArena WebDev | 1339 | 1517 |
| LMArena Agent | – | -3.1 |
| LiveBench | 74.0% | 73.1% |
| LiveBench Reasoning | 77.7% | 83.3% |
| LiveBench Coding | 83.6% | 74.2% |
| LiveBench Agentic Coding | 49.4% | 43.6% |
| LiveBench Mathematics | 88.8% | 85.3% |
| LiveBench Data Analysis | 78.2% | 71.8% |
| LiveBench Language | 73.7% | 79.7% |
| AA Intelligence Index | 28.5 | 29.9 |
| IFBench | 77.6% | 80.5% |
| AA-LCR | 82.3% | 79.0% |
| MMMU-Pro | 76.3% | – |
| AA-Omniscience | -2.2 | 13.5 |
| Terminal-Bench Hard | 37.1% | 50.8% |
| GPQA Diamond (AA) | 89.9% | 92.3% |
| Humanity's Last Exam (AA) | 35.7% | 40.5% |
| SciCode (AA) | – | 49.5% |
| τ²-Bench Telecom (AA) | 92.1% | 94.7% |
| LiveCodeBench | 88.0% | 87.1% |
| MMLU-Pro | – | 89.3% |
| IOI | – | 46.8% |
| LegalBench | – | 84.9% |
| CorpFin | – | 63.7% |
| TaxEval | – | 75.3% |
| Terminal-Bench 2.1 (Vals) | – | 61.0% |
| SWE-bench (Vals) | – | 68.8% |
| GPQA Diamond (Vals) | – | 90.2% |
| Vals Index | – | 44.8 |
| SWE-Bench Pro | 41.0% | – |
| EQ-Bench 4 | – | 1110 |
| SWE-bench Verified (bash only) | 72.8% | – |
| SWE-bench Verified (any scaffold) | 72.8% | – |
Data as of 2026-09-10. Best configuration of each model; every score links to its source on the model pages.
GPT-5.2-Codex vs Qwen3 7: questions
- Is GPT-5.2-Codex better than Qwen3 7?
- GPT-5.2-Codex and Qwen3 7 (max) are level on quality (60.8 vs 61.3). The BenchLeader Index combines every independent quality benchmark; Qwen3 7 (max) is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is GPT-5.2-Codex better than Qwen3 7 for coding?
- Qwen3 7 scores higher in coding (61 vs 56 on the category index, where 50 is average).
- Is GPT-5.2-Codex better than Qwen3 7 for agentic tasks?
- GPT-5.2-Codex scores higher in agentic tasks (64 vs 56 on the category index, where 50 is average).
- Which is cheaper, GPT-5.2-Codex or Qwen3 7?
- Qwen3 7 is cheaper: $3.75 against $4.81 per million tokens, blended at three input tokens per output token.
- Which is faster, GPT-5.2-Codex or Qwen3 7?
- Qwen3 7 streams faster: 169 against 25 output tokens per second.
- Which has the larger context window?
- Qwen3 7 accepts more context: 1M against 400k tokens.