DeepSeek V4 Pro vs Qwen3 8
DeepSeek V4 Pro 
Qwen3 8 Verdict
- Qwen3 8 (max) leads on quality: 66.5 vs 64.2.
- DeepSeek V4 Pro (max) is stronger in agents & tools, composite, instruction following.
- Qwen3 8 (max) is stronger in coding, knowledge, maths, reasoning, human preference, multimodal.
- DeepSeek V4 Pro (max) is 4.9× cheaper ($0.544 vs $2.67 per 1M blended).
- DeepSeek V4 Pro (max) streams 2.1× faster (81 vs 39 tokens per second).
| Metric | DeepSeek V4 Pro (max) | Qwen3 8 (max) |
|---|---|---|
| BenchLeader Index | 64.2 | 66.5 |
| Agents & tools score | 70.2 | 68.3 |
| Coding score | 58.8 | 66.5 |
| Composite score | 75.0 | 74.3 |
| Instruction following score | 74.3 | – |
| Knowledge score | 58.1 | 63.5 |
| Long context score | 66.6 | 66.6 |
| Maths score | 53.7 | 67.0 |
| Reasoning score | 70.9 | 76.3 |
| Human preference score | – | 68.2 |
| Multimodal score | – | 67.4 |
| Blended price $/M | $0.544 | $2.67 |
| Output speed | 81 tok/s | 39 tok/s |
| Time to first answer | 55.6 s | 55.1 s |
| Context window | 1M | 1M |
| GPQA Diamond | 89.7% | – |
| FrontierMath Tiers 1–3 | 45.3% | – |
| FrontierMath Tier 4 | 2.4% | – |
| OTIS Mock AIME | 96.7% | – |
| SWE-bench Verified (Epoch) | 77.6% | – |
| SimpleQA Verified | 47.0% | – |
| SciCode | 50.0% | 52.9% |
| WeirdML | 48.9% | – |
| ProofBench | 16.0% | 58.0% |
| Epoch Capabilities Index | – | 156.6 |
| LMArena Text | – | 1481 |
| LMArena Hard Prompts | – | 1503 |
| LMArena Coding | – | 1522 |
| LMArena WebDev | – | 1671 |
| LMArena Vision | – | 1315 |
| LMArena Agent | – | 3.3 |
| LiveBench | – | 78.5% |
| LiveBench Reasoning | – | 88.2% |
| LiveBench Coding | – | 72.9% |
| LiveBench Agentic Coding | – | 64.7% |
| LiveBench Mathematics | – | 91.3% |
| LiveBench Data Analysis | – | 78.4% |
| LiveBench Language | – | 79.7% |
| LiveBench Instruction Following | – | 74.1% |
| AA Intelligence Index | 36.3 | 45.4 |
| IFBench | 76.5% | – |
| AA-LCR | 80.3% | 80.3% |
| MMMU-Pro | – | 82.8% |
| AA-Omniscience | 0.8 | 12.0 |
| Terminal-Bench Hard | 46.2% | – |
| GPQA Diamond (AA) | 92.8% | 92.8% |
| Humanity's Last Exam (AA) | 41.0% | 43.1% |
| SciCode (AA) | 51.0% | 53.2% |
| τ²-Bench Telecom (AA) | 96.2% | – |
| LiveCodeBench | 87.5% | 87.8% |
| MMLU-Pro | 87.3% | 88.6% |
| IOI | – | 68.9% |
| LegalBench | 80.3% | 83.6% |
| CorpFin | 61.4% | 65.8% |
| TaxEval | 72.1% | 75.5% |
| Terminal-Bench 2.1 (Vals) | – | 67.4% |
| SWE-bench (Vals) | 77.4% | 85.6% |
| GPQA Diamond (Vals) | 89.4% | 93.7% |
| Vals Index | 42.9 | 51.8 |
| AIME 2026 | 96.7% | – |
| HMMT February 2026 | 93.9% | – |
| MathArena Apex | 28.1% | – |
| CritPt | 18.0% | 20.0% |
| GDPval (AA) | 49.7% | 58.2% |
| τ²-Bench Banking (AA) | 39.6% | 51.3% |
| ITBench SRE (AA) | 38.3% | – |
| Analyst Agent (AA) | 18.8% | – |
| APEX-Agents (AA) | 24.3% | – |
| CaseLaw v2 | 59.4% | – |
| Code Migration | 26.2% | 24.0% |
| Excel Modeling Benchmark | 51.6% | 60.1% |
| Finance Agent v2 | 44.1% | 50.6% |
| Harvey's Legal Agent Benchmark | 3.8% | 10.4% |
| Legal Research Bench | 23.1% | 47.6% |
| MedCode | 40.5% | 40.7% |
| MedScribe | 75.1% | 85.0% |
| MMMU-Pro (Vals) | – | 88.0% |
| MortgageTax | – | 64.0% |
| MysteryMechanism | – | 23.9% |
| ProgramBench | 0.0% | 0.0% |
| Public Benefits Bench | 62.9% | 67.1% |
| SAGE | – | 51.3% |
| SkillsBench | 51.3% | 42.0% |
| Tax Agent Bench | – | 66.0% |
| Terminal-Bench 4.0 (Vals) | – | 24.8% |
| Terminal-Bench Science | – | 1.4% |
| Vals Multimodal Index | – | 65.4% |
| Vibe Code Bench 1-100 | – | 12.8% |
| Vibe Code Bench v1.1 | 49.9% | 64.7% |
| LMArena Maths | – | 1498 |
| LMArena Creative Writing | – | 1472 |
| LMArena Instruction Following | – | 1477 |
| LMArena Multi-turn | – | 1495 |
| LMArena Longer Queries | – | 1492 |
| Chess Puzzles | 20.0% | – |
| LMCA | 41.2% | – |
| DTBench | 90.7% | – |
Data as of 2026-09-19. Best configuration of each model; every score links to its source on the model pages.
DeepSeek V4 Pro vs Qwen3 8: questions
- Is DeepSeek V4 Pro better than Qwen3 8?
- Qwen3 8 (max) leads on quality: 66.5 vs 64.2. The BenchLeader Index combines every independent quality benchmark; Qwen3 8 (max) is ahead overall as of 2026-09-19, but check the category scores for your use.
- Is DeepSeek V4 Pro better than Qwen3 8 for coding?
- Qwen3 8 scores higher in coding (67 vs 59 on the category index, where 50 is average).
- Is DeepSeek V4 Pro better than Qwen3 8 for agentic tasks?
- DeepSeek V4 Pro scores higher in agentic tasks (70 vs 68 on the category index, where 50 is average).
- Which is cheaper, DeepSeek V4 Pro or Qwen3 8?
- DeepSeek V4 Pro is cheaper: $0.544 against $2.67 per million tokens, blended at three input tokens per output token.
- Which is faster, DeepSeek V4 Pro or Qwen3 8?
- DeepSeek V4 Pro streams faster: 81 against 39 output tokens per second.
- Which has the larger context window?
- Both accept 1M tokens of context.