Qwen3.8 Max
| Released | 2026-08-03 |
|---|---|
| Price per 1M | $2.00 in · $6.00 out |
| Overall score | 59.9 · rank 30 |
| Context window | 1,000,000 tokens |
| Provider | Alibaba |
| Weights | Proprietary |
Scores
| Dimension | Score | Rank | Benchmarks |
|---|---|---|---|
| Overall: Balanced | 59.9 | 30 | 45 |
| Overall: Coding | 59.3 | 37 | 33 |
| Overall: Reasoning | 60.8 | 29 | 15 |
| Overall: Agentic work | 60.3 | 30 | 30 |
| Overall: Frontend | 59.0 | 32 | 15 |
| Coding | 58.6 | 39 | 10 |
| Agents | 60.6 | 23 | 12 |
| Reasoning | 61.9 | 28 | 8 |
| Math | 60.7 | 25 | 4 |
| Knowledge | 58.3 | 40 | 3 |
| Frontend | 58.8 | 32 | 3 |
| General | 59.4 | 33 | 3 |
| Multimodal | 61.4 | 10 | 2 |
| Value vs price | +8.9 | 37 |
Configurations
| Configuration | Balanced | Benchmarks |
|---|---|---|
| Unspecified setting | 59.8 ±0.8 | 35 |
| Extra High | 60.8 ±2.0 | 11 |
Evidence
Coding 58.6 ±0.9
| Benchmark | Result | Source | Date |
|---|---|---|---|
| FrontierSWE | 15.8% | Epoch AI | 2026-11-11 |
| DeepSWE | 57.5% | Epoch AI | 2026-11-11 |
| LiveBench 2.0 Coding | 60.9% | LiveBench | 2026-11-06 |
| MirrorCode | 52.2% | Epoch AI | 2026-10-21 |
| IOI | 68.9% | Vals AI | 2026-10-29 |
| SWE-bench Multimodal | 42.5% | Vals AI | 2026-11-09 |
| SWE-bench Verified (Vals run) | 85.6% | Vals AI | 2026-09-01 |
| SciCode | 53.2% | Epoch AI | 2026-11-11 |
| LiveCodeBench (Vals run) | 87.9% | Vals AI | 2026-09-01 |
| ProgramBench | 0.0% | Vals AI | 2026-11-09 |
Agents 60.6 ±1.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Terminal-Bench 4.0 (Vals run) | 34.3% | Vals AI | 2026-11-09 |
| Terminal-Bench-Science | 12.9% | Vals AI | 2026-11-09 |
| APEX-Agents | 63.3% | Epoch AI | 2026-11-11 |
| LiveBench 2.0 Agentic | 26.6% | LiveBench | 2026-11-06 |
| LiveBench 2.0 Browser Use | 49.4% | LiveBench | 2026-11-06 |
| Terminal-Bench 2.1 (Vals run) | 67.4% | Vals AI | 2026-10-01 |
| tau2-bench Banking Knowledge | 55.2% | tau2-bench | 2026-08-03 |
| EMB v1.1 (Excel financial modeling) | 66.0% | Vals AI | 2026-11-09 |
| LiveBench 2.0 Research | 72.8% | LiveBench | 2026-11-06 |
| Tax Agent Bench | 32.1% | Vals AI | 2026-11-09 |
| SkillsBench | 42.0% | Vals AI | 2026-10-29 |
| CyberBench v1.1 | 28.6% | Vals AI | 2026-11-05 |
Reasoning 61.9 ±2.0
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Extended NYT Connections | 88.3% | Lech Mazur | 2026-11-10 |
| MysteryMechanism | 23.9% | Vals AI | 2026-11-09 |
| CritPt | 20.0% | Epoch AI | 2026-11-11 |
| GPQA Diamond | 92.7% | Epoch AI | 2026-08-04 |
| GPQA Diamond (Vals run) | 93.7% | Vals AI | 2026-09-01 |
| Chess Puzzles | 46.0% | Epoch AI | 2026-10-13 |
| LiveBench 2.0 Reasoning | 88.2% | LiveBench | 2026-11-06 |
| Mystery Game Puzzles | 38.0% | Epoch AI | 2026-08-05 |
| LiveBench 2.0 Data Analysis | 78.4% | LiveBench | 2026-11-06 |
Math 60.7 ±2.1
| Benchmark | Result | Source | Date |
|---|---|---|---|
| FrontierMath Tiers 1-3 | 74.7% | Epoch AI | 2026-08-04 |
| FrontierMath Tier 4 | 46.3% | Epoch AI | 2026-08-04 |
| OTIS Mock AIME 2024-2025 | 99.4% | Epoch AI | 2026-08-04 |
| ProofBench v1.1 | 58.0% | Vals AI | 2026-11-09 |
| ProofBench | 58.0% | Epoch AI | 2026-11-11 |
Knowledge 58.3 ±3.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| SimpleQA Verified | 45.8% | Epoch AI | 2026-08-27 |
| MMLU-Pro (Vals run) | 88.6% | Vals AI | 2026-09-01 |
| Legal Research Bench | 47.6% | Vals AI | 2026-11-09 |
Frontend 58.8 ±3.2
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LiveBench 2.0 Design | 76.6% | LiveBench | 2026-11-06 |
| Vibe Code Bench v1.1 | 64.7% | Vals AI | 2026-11-09 |
| LMArena WebDev | 1,670 | LMArena | 2026-11-09 |
General 59.4 ±2.8
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LMArena Text | 1,484 | LMArena | 2026-11-09 |
| LLM Debate Benchmark | 1,546 | Lech Mazur | 2026-10-26 |
| Creative Story-Writing (Lech Mazur) | 52.0% | Lech Mazur | 2026-10-23 |
Multimodal 61.4 ±2.3
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MMMU-Pro (Vals run) | 88.0% | Vals AI | 2026-09-01 |
| LMArena Vision | 1,299 | LMArena | 2026-11-09 |
Names matched to this model
qwen-3-8-max, qwen3-8-max