Qwen3-Max-Instruct: benchmark scores, price and ranking
Qwen3-Max-Instruct from Alibaba ranks #377 of 743 LLMs on the LLMs Tiger aggregate benchmark score (48.2), based on 12 benchmark results captured up to 2026-10-13.
- Organisation
- Alibaba
- Type
- Frontier model
- Open weights
- No
- Released
- 2025-09-24 (source: epoch)
- Aggregate rank
- #377 of 743 (score 48.2)
Relative to other models, Qwen3-Max-Instruct is strongest on SimpleQA Verified (#21 of 69) and weakest on Mystery Games (#57 of 59).
Across the 12 benchmarks where at least ten models are scored, Qwen3-Max-Instruct is in the top 10% on 0 and in the top half on 5.
Qwen3-Max-Instruct benchmark results
| Benchmark | Score | Rank | Source | Captured |
|---|---|---|---|---|
| SimpleQA Verified | 48.7% | #21 of 69 | Epoch AI Benchmarking Hub | 2026-10-13 |
| DTBench | 82.1% | #45 of 146 | Epoch AI Benchmarking Hub | 2026-10-13 |
| OTIS Mock AIME | 73.3% | #73 of 189 | Epoch AI Benchmarking Hub | 2026-10-13 |
| Fiction.liveBench | 66.7% | #19 of 47 | Epoch AI Benchmarking Hub | 2026-10-13 |
| GPQA Diamond | 72.6% | #136 of 275 | Epoch AI Benchmarking Hub | 2026-10-13 |
| LMCA | 28.3% | #59 of 108 | Epoch AI Benchmarking Hub | 2026-10-13 |
| Chess Puzzles | 4.0% | #87 of 122 | Epoch AI Benchmarking Hub | 2026-10-13 |
| CL-bench | 14.5% | #13 of 16 | Epoch AI Benchmarking Hub | 2026-10-13 |
| ALE-Bench | 370 | #65 of 79 | Epoch AI Benchmarking Hub | 2026-10-13 |
| FrontierMath T1-3 | 18.9% | #57 of 68 | Epoch AI Benchmarking Hub | 2026-10-13 |
| Vending-Bench 2 | $72 | #18 of 19 | Epoch AI Benchmarking Hub | 2026-10-13 |
| Mystery Games | 5.0% | #57 of 59 | Epoch AI Benchmarking Hub | 2026-10-13 |
About Qwen3-Max-Instruct
Qwen3-Max-Instruct is a frontier model from Alibaba. It was released on 2025-09-24, according to Epoch AI. No API price is listed for it yet. We hold 12 benchmark results for it across 4 areas, and none is in the top tenth of its board.
Where Qwen3-Max-Instruct stands
On our aggregate score, Qwen3-Max-Instruct is #377 of 743 ranked models, around the middle of the table. Its score is 48.2; the scale is centred on 50, so it sits 1.8 points below the typical model. That beats 49% of the ranked field. Among the 74 ranked Alibaba models it is #38. The next one up is qwen2:72b-instruct-q8_0 at 48.2. The next one down is qwen3-coder-plus at 48.0.
Qwen3-Max-Instruct, benchmark by benchmark
SimpleQA Verified puts Qwen3-Max-Instruct at 48.7%, #21 of 69. That beats 70% of the models tested, but 27.3 points short of gemini-4-argon-high (76.0%). The results either side are Grok 4.6 (xhigh) (48.9%) and Grok 4.5 (high) (48.3%). It covers short factual questions. The board median is 41.0%, so Qwen3-Max-Instruct is 7.7 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
On DTBench, Qwen3-Max-Instruct reaches 82.1%. It ranks #45 of 146, ahead of 69% of the field, and 16.8 points behind Claude Opus 5.5 (max) at 98.9%. The results either side are Qwen3.5 27B (82.4%) and Qwen 3.6 Plus (2026-04-02) (81.9%). It covers dynamic tool-use benchmark. The board median is 68.8%, so Qwen3-Max-Instruct is 13.3 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Qwen3-Max-Instruct manages 73.3% on OTIS Mock AIME, good for #73 of 189. Claude Fable 5.1 (max) leads at 100.0%, 26.7 points further up, and 61% of models sit below. The results either side are o1 (high) (73.3%) and Gemini 2.5 Flash Preview (Apr 2025) (73.1%). It covers competition maths (AIME 2024-25 style). The board median is 57.8%, so Qwen3-Max-Instruct is 15.6 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Fiction.liveBench puts Qwen3-Max-Instruct at 66.7%, #19 of 47. That beats 60% of the models tested, but 27.7 points short of grok-4-0709 (94.4%). The results either side are qwen-max-2025-01-25 (66.7%) and GPT-4.1 (63.9%). It covers long-context story comprehension (16k). The board median is 61.1%, so Qwen3-Max-Instruct is 5.6 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
On GPQA Diamond, Qwen3-Max-Instruct reaches 72.6%. It ranks #136 of 275, ahead of 51% of the field, and 23.2 points behind GPT-6 Astra (max) at 95.8%. The results either side are Gemma 4 26B A4B (73.2%) and qwen3-30b-a3b-thinking-2507 (72.2%). It covers expert-level science questions. The board median is 71.7%, so Qwen3-Max-Instruct is 0.9 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Qwen3-Max-Instruct lands below the middle on LMCA: 28.3%, #59 of 108, with 54% of the field scoring higher. Claude Opus 5.5 (max) leads at 68.2%; the closest result above is DeepSeek-V3.1-Terminus at 28.6%. It covers language-model capability assessment. The board median is 29.2%, so Qwen3-Max-Instruct is 1.0 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
ALE-Bench is a weak spot. Qwen3-Max-Instruct scores 370, #65 of 79, behind 81% of the models measured. The leader, GPT-6 Astra (max), reaches 2,951. It covers AtCoder heuristic-optimisation contests (performance rating). The board median is 764, so Qwen3-Max-Instruct is 394 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
FrontierMath T1-3 is a weak spot. Qwen3-Max-Instruct scores 18.9%, #57 of 68, behind 82% of the models measured. The leader, GPT-6 Astra (max), reaches 93.7%. It covers research-level maths, tiers 1-3. The board median is 55.6%, so Qwen3-Max-Instruct is 36.7 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Vending-Bench 2 is a weak spot. Qwen3-Max-Instruct scores $72, #18 of 19, behind 89% of the models measured. The leader, Kimi K2.6, reaches $6,205. It covers running a vending business for a year (final money, $). The board median is $2,377, so Qwen3-Max-Instruct is 2,305 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Mystery Games is a weak spot. Qwen3-Max-Instruct scores 5.0%, #57 of 59, behind 95% of the models measured. The leader, GPT-6 Astra (max), reaches 84.0%. It covers deductive reasoning puzzles. The board median is 23.0%, so Qwen3-Max-Instruct is 18.0 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
The full table above also lists the other 2 results, including boards with fewer than ten models, where a rank says little.
Strengths and weaknesses
Reasoning, maths and science: uneven, ahead of 43% of the field on average across 3 benchmarks. Its best board here is OTIS Mock AIME (#73 of 189) and its weakest is FrontierMath T1-3 (#57 of 68). General knowledge and chat: uneven, ahead of 33% of the field on average across 7 benchmarks. Its best board here is SimpleQA Verified (#21 of 69) and its weakest is Mystery Games (#57 of 59).
Release history
The Alibaba model released before it was qwen3-coder-plus on 2025-09-23, which scores 48.0. The next one was qwen3-coder-next on 2026-02-02, at 45.5.
Reading these numbers
The 12 scores on this page come from 1 source: Epoch AI Benchmarking Hub (12). No official leaderboard supplied any of them, so treat them as third-party measurements.
Models ranked nearby
- #374 Qwen2.5 (7B + 72B) (48.2)
- #375 claude-3-7-sonnet-20250219 (48.2)
- #376 qwen2:72b-instruct-q8_0 (48.2)
- #378 internlm3-8b-instruct (48.2)
- #379 kimi-k2.6-thinking (48.2)
- #380 o1 (high) (48.1)
More from Alibaba
- Qwen 3.6 Max (Preview) (#25, 54.1)
- Qwen3.8 Max (xhigh) (#37, 53.2)
- Qwen3.7 Max (#71, 52.3)
- ollama/qwen2.5-coder:32b (#93, 51.9)
- Qwen3 235B A22B diff, no think, Alibaba API (#122, 51.5)
- Qwen2.5-Coder-14B-Instruct (#135, 51.3)
Compare it on the live leaderboard → · How the aggregate score works