GPT-5.2 vs Qwen3 7
GPT-5.2 
Qwen3 7 Verdict
- GPT-5.2 and Qwen3 7 (max) are level on quality (61.6 vs 61.3).
- GPT-5.2 is stronger in agents & tools, composite, human preference, long context, multimodal.
- Qwen3 7 (max) is stronger in coding, instruction following, knowledge, reasoning, maths.
- Qwen3 7 (max) is 1.3× cheaper ($3.75 vs $4.81 per 1M blended).
- Qwen3 7 (max) streams 2.5× faster (169 vs 68 tokens per second).
| Metric | GPT-5.2 | Qwen3 7 (max) |
|---|---|---|
| BenchLeader Index | 61.6 | 61.3 |
| Agents & tools score | 61.2 | 56.0 |
| Coding score | 53.7 | 61.4 |
| Composite score | 67.5 | 56.4 |
| Human preference score | 67.9 | 57.2 |
| Instruction following score | 67.4 | 77.6 |
| Knowledge score | 61.1 | 63.6 |
| Long context score | 67.9 | 66.0 |
| Multimodal score | 59.9 | – |
| Reasoning score | 61.3 | 66.8 |
| Maths score | – | 57.4 |
| Blended price $/M | $4.81 | $3.75 |
| Output speed | 68 tok/s | 169 tok/s |
| Time to first answer | 129.4 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% |
| Humanity's Last Exam | 27.8% | – |
| Terminal-Bench | 64.9% | – |
| SimpleBench | 45.8% | 70.4% |
| SciCode | – | 48.8% |
| Remote Labor Index | 2.1% | – |
| APEX-Agents | 23.0% | – |
| ProofBench | – | 26.0% |
| Epoch Capabilities Index | 153.5 | 153.7 |
| LMArena Text | 1476 | 1474 |
| LMArena Hard Prompts | 1497 | 1495 |
| LMArena Coding | 1515 | 1525 |
| LMArena WebDev | 1417 | 1517 |
| LMArena Vision | 1268 | – |
| LMArena Agent | – | -3.1 |
| LiveBench | – | 73.1% |
| LiveBench Reasoning | – | 83.3% |
| LiveBench Coding | – | 74.2% |
| LiveBench Agentic Coding | – | 43.6% |
| LiveBench Mathematics | – | 85.3% |
| LiveBench Data Analysis | – | 71.8% |
| LiveBench Language | – | 79.7% |
| AA Intelligence Index | 30.4 | 29.9 |
| IFBench | 75.4% | 80.5% |
| AA-LCR | 82.7% | 79.0% |
| AA-Omniscience | -0.9 | 13.5 |
| Terminal-Bench Hard | 47.0% | 50.8% |
| GPQA Diamond (AA) | 90.3% | 92.3% |
| Humanity's Last Exam (AA) | 37.7% | 40.5% |
| SciCode (AA) | – | 49.5% |
| τ²-Bench Telecom (AA) | 84.8% | 94.7% |
| LiveCodeBench | – | 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 | 29.9% | – |
| VISTA | 46.6% | – |
| MultiNRC | 42.2% | – |
| TutorBench | 53.5% | – |
| EQ-Bench 4 | – | 1110 |
| Kagi LLM Benchmark | 73.3% | – |
| SWE-bench Verified (bash only) | 69.0% | – |
| ARC-AGI-1 | 94.5% | – |
| ARC-AGI-2 | 72.9% | – |
| 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 Qwen3 7: questions
- Is GPT-5.2 better than Qwen3 7?
- GPT-5.2 and Qwen3 7 (max) are level on quality (61.6 vs 61.3). The BenchLeader Index combines every independent quality benchmark; GPT-5.2 is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is GPT-5.2 better than Qwen3 7 for coding?
- Qwen3 7 scores higher in coding (61 vs 54 on the category index, where 50 is average).
- Is GPT-5.2 better than Qwen3 7 for agentic tasks?
- GPT-5.2 scores higher in agentic tasks (61 vs 56 on the category index, where 50 is average).
- Which is cheaper, GPT-5.2 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 or Qwen3 7?
- Qwen3 7 streams faster: 169 against 68 output tokens per second.
- Which has the larger context window?
- Qwen3 7 accepts more context: 1M against 400k tokens.