Qwen3 7 vs Qwen3 8
Qwen3 7 Verdict
- Qwen3 8 (max) leads on quality: 63.6 vs 61.3.
- Qwen3 7 (max) is stronger in instruction following, knowledge, long context.
- Qwen3 8 (max) is stronger in coding, composite, human preference, maths, reasoning, multimodal.
- Qwen3 8 (max) is 1.4× cheaper ($2.67 vs $3.75 per 1M blended).
- Qwen3 7 (max) streams 4.5× faster (169 vs 38 tokens per second).
| Metric | Qwen3 7 (max) | Qwen3 8 (max) |
|---|---|---|
| BenchLeader Index | 61.3 | 63.6 |
| Agents & tools score | 56.0 | 56.0 |
| Coding score | 61.4 | 69.7 |
| Composite score | 56.4 | 70.9 |
| Human preference score | 57.2 | 68.3 |
| Instruction following score | 77.6 | – |
| Knowledge score | 63.6 | 62.3 |
| Long context score | 66.0 | 65.7 |
| Maths score | 57.4 | 62.4 |
| Reasoning score | 66.8 | 68.1 |
| Multimodal score | – | 67.2 |
| Blended price $/M | $3.75 | $2.67 |
| Output speed | 169 tok/s | 38 tok/s |
| Time to first answer | 16.5 s | 55.3 s |
| Context window | 1M | 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% | – |
| SimpleBench | 70.4% | – |
| SciCode | 48.8% | 52.9% |
| ProofBench | 26.0% | 58.0% |
| Epoch Capabilities Index | 153.7 | 156.6 |
| LMArena Text | 1474 | 1480 |
| LMArena Hard Prompts | 1495 | 1502 |
| LMArena Coding | 1525 | 1520 |
| LMArena WebDev | 1517 | 1670 |
| LMArena Vision | – | 1313 |
| LMArena Agent | -3.1 | 3.9 |
| LiveBench | 73.1% | 78.5% |
| LiveBench Reasoning | 83.3% | 88.2% |
| LiveBench Coding | 74.2% | 72.9% |
| LiveBench Agentic Coding | 43.6% | 64.7% |
| LiveBench Mathematics | 85.3% | 91.3% |
| LiveBench Data Analysis | 71.8% | 78.4% |
| LiveBench Language | 79.7% | 79.7% |
| AA Intelligence Index | 29.9 | 40.3 |
| IFBench | 80.5% | – |
| AA-LCR | 79.0% | 78.3% |
| MMMU-Pro | – | 82.3% |
| AA-Omniscience | 13.5 | 3.4 |
| Terminal-Bench Hard | 50.8% | – |
| GPQA Diamond (AA) | 92.3% | 92.7% |
| Humanity's Last Exam (AA) | 40.5% | 43.0% |
| SciCode (AA) | 49.5% | 53.2% |
| τ²-Bench Telecom (AA) | 94.7% | – |
| LiveCodeBench | 87.1% | 87.8% |
| MMLU-Pro | 89.3% | 88.6% |
| IOI | 46.8% | 73.0% |
| LegalBench | 84.9% | 83.6% |
| CorpFin | 63.7% | 65.8% |
| TaxEval | 75.3% | 75.5% |
| Terminal-Bench 2.1 (Vals) | 61.0% | 67.4% |
| SWE-bench (Vals) | 68.8% | 85.6% |
| GPQA Diamond (Vals) | 90.2% | 93.7% |
| Vals Index | 44.8 | 51.8 |
| EQ-Bench 4 | 1110 | – |
Data as of 2026-09-10. Best configuration of each model; every score links to its source on the model pages.
Qwen3 7 vs Qwen3 8: questions
- Is Qwen3 7 better than Qwen3 8?
- Qwen3 8 (max) leads on quality: 63.6 vs 61.3. The BenchLeader Index combines every independent quality benchmark; Qwen3 8 (max) is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is Qwen3 7 better than Qwen3 8 for coding?
- Qwen3 8 scores higher in coding (70 vs 61 on the category index, where 50 is average).
- Is Qwen3 7 better than Qwen3 8 for agentic tasks?
- They are level in agentic tasks (56 each).
- Which is cheaper, Qwen3 7 or Qwen3 8?
- Qwen3 8 is cheaper: $2.67 against $3.75 per million tokens, blended at three input tokens per output token.
- Which is faster, Qwen3 7 or Qwen3 8?
- Qwen3 7 streams faster: 169 against 38 output tokens per second.
- Which has the larger context window?
- Both accept 1M tokens of context.