qwen3-30b-a3b-thinking-2507: benchmark scores, price and ranking
qwen3-30b-a3b-thinking-2507 from Alibaba ranks #456 of 743 LLMs on the LLMs Tiger aggregate benchmark score (47.2), based on 8 benchmark results captured up to 2026-10-13.
- Organisation
- Alibaba
- Type
- Open-weight / local model
- Open weights
- Yes
- Cheapest API price
- $0.200 in / $2.40 out per million tokens (via openrouter, observed 2026-10-09T15:47Z)
- Aggregate rank
- #456 of 743 (score 47.2)
Relative to other models, qwen3-30b-a3b-thinking-2507 is strongest on MMLU-Pro (#34 of 136) and weakest on SciCode (#87 of 106).
Across the 8 benchmarks where at least ten models are scored, qwen3-30b-a3b-thinking-2507 is in the top 10% on 0 and in the top half on 4.
Its cheapest listed input price, $0.200 per million tokens, is lower than that of 67% of the 427 models with a listed price.
qwen3-30b-a3b-thinking-2507 benchmark results
| Benchmark | Score | Rank | Source | Captured |
|---|---|---|---|---|
| MMLU-Pro | 80.9% | #34 of 136 | HuggingFace MMLU-Pro leaderboard | 2026-10-13 |
| OTIS Mock AIME | 70.3% | #79 of 189 | Epoch AI Benchmarking Hub | 2026-10-13 |
| DTBench | 69.3% | #72 of 146 | Epoch AI Benchmarking Hub | 2026-10-13 |
| GPQA Diamond | 72.2% | #137 of 275 | HuggingFace GPQA Diamond leaderboard | 2026-10-13 |
| Chess Puzzles | 8.0% | #69 of 122 | Epoch AI Benchmarking Hub | 2026-10-13 |
| LMCA | 19.5% | #73 of 108 | Epoch AI Benchmarking Hub | 2026-10-13 |
| CritPt | 0.3% | #81 of 115 | Epoch AI Benchmarking Hub | 2026-10-13 |
| SciCode | 33.3% | #87 of 106 | Epoch AI Benchmarking Hub | 2026-10-13 |
About qwen3-30b-a3b-thinking-2507
qwen3-30b-a3b-thinking-2507 is an open-weight model that can run on your own hardware from Alibaba. The cheapest API listing we track charges $0.200 per million input tokens and $2.40 per million output tokens, through openrouter. We hold 8 benchmark results for it across 4 areas, and none is in the top tenth of its board.
Where qwen3-30b-a3b-thinking-2507 stands
On our aggregate score, qwen3-30b-a3b-thinking-2507 is #456 of 743 ranked models, in the lower half. Its score is 47.2; the scale is centred on 50, so it sits 2.8 points below the typical model. That beats 39% of the ranked field. Among the 74 ranked Alibaba models it is #44. The next one up is Qwen3-235b-a22b Instruct (25/07) at 47.5. The next one down is Qwen 3.6 Flash (2026-04-16) at 47.2.
qwen3-30b-a3b-thinking-2507, benchmark by benchmark
MMLU-Pro puts qwen3-30b-a3b-thinking-2507 at 80.9%, #34 of 136. That beats 75% of the models tested, but 7.2 points short of Darwin-180B-RSI (88.1%). The results either side are Kimi K2 Instruct (81.0%) and MiniMax-M2.5 (80.1%). It covers 57-subject knowledge + reasoning. The board median is 55.3%, so qwen3-30b-a3b-thinking-2507 is 25.6 points above it. Source: HuggingFace MMLU-Pro leaderboard, captured 2026-10-13.
On OTIS Mock AIME, qwen3-30b-a3b-thinking-2507 reaches 70.3%. It ranks #79 of 189, ahead of 58% of the field, and 29.7 points behind Claude Fable 5.1 (max) at 100.0%. The results either side are gemini-2.5-flash-preview-05-20 (70.8%) and claude-opus-4-1-20250805_27K (68.9%). It covers competition maths (AIME 2024-25 style). The board median is 57.8%, so qwen3-30b-a3b-thinking-2507 is 12.5 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
qwen3-30b-a3b-thinking-2507 manages 69.3% on DTBench, good for #72 of 146. Claude Opus 5.5 (max) leads at 98.9%, 29.6 points further up, and 51% of models sit below. The results either side are Mistral Small 4 (70.9%) and GPT-4.1 mini (68.8%). It covers dynamic tool-use benchmark. The board median is 68.8%, so qwen3-30b-a3b-thinking-2507 is 0.5 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
GPQA Diamond puts qwen3-30b-a3b-thinking-2507 at 72.2%, #137 of 275. That beats 50% of the models tested, but 23.5 points short of GPT-6 Astra (max) (95.8%). The results either side are Qwen3-Max-Instruct (72.6%) and GLM-4.5-Air (71.7%). It covers expert-level science questions. The board median is 71.7%, so qwen3-30b-a3b-thinking-2507 is 0.5 points above it. Source: HuggingFace GPQA Diamond leaderboard, captured 2026-10-13.
qwen3-30b-a3b-thinking-2507 lands below the middle on Chess Puzzles: 8.0%, #69 of 122, with 56% of the field scoring higher. GPT-6 Astra (max) leads at 72.0%; the closest result above is claude-haiku-4-5-20251001_32K at 8.0%. It covers chess tactics puzzles. The board median is 12.5%, so qwen3-30b-a3b-thinking-2507 is 4.5 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
qwen3-30b-a3b-thinking-2507 lands below the middle on LMCA: 19.5%, #73 of 108, with 67% of the field scoring higher. Claude Opus 5.5 (max) leads at 68.2%; the closest result above is Mistral Small 4 at 20.6%. It covers language-model capability assessment. The board median is 29.2%, so qwen3-30b-a3b-thinking-2507 is 9.7 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
qwen3-30b-a3b-thinking-2507 lands below the middle on CritPt: 0.3%, #81 of 115, with 70% of the field scoring higher. GPT-5.6 Sol (max) leads at 32.3%; the closest result above is o3-mini (high) at 0.3%. It covers unpublished research-level physics problems. The board median is 2.9%, so qwen3-30b-a3b-thinking-2507 is 2.6 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
SciCode is a weak spot. qwen3-30b-a3b-thinking-2507 scores 33.3%, #87 of 106, behind 81% of the models measured. The leader, Claude Opus 5.5 (max), reaches 66.9%. It covers scientific research code. The board median is 44.3%, so qwen3-30b-a3b-thinking-2507 is 11.0 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Strengths and weaknesses
Reasoning, maths and science: solid, ahead of 53% of the field on average across 4 benchmarks. Its best board here is MMLU-Pro (#34 of 136) and its weakest is CritPt (#81 of 115). General knowledge and chat: uneven, ahead of 38% of the field on average across 2 benchmarks. Its best board here is Chess Puzzles (#69 of 122) and its weakest is LMCA (#73 of 108).
Price and value
The median input price across the 427 priced models is $0.500 per million tokens, so qwen3-30b-a3b-thinking-2507 costs less than the typical listing. Models that score within 1.5 points or better and cost less: Qwen2.5-Coder-7B-Instruct ($0.010, 49.5), DeepSeek V4.1 Flash (high) ($0.010, 48.9), Qwen2.5-Coder-3B-Instruct ($0.010, 46.5). Similar scores cost more on DeepSeek-V3.1 ($0.250) and gemini-3.1-flash-lite (high) ($0.250).
Reading these numbers
The 8 scores on this page come from 3 sources: Epoch AI Benchmarking Hub (6), HuggingFace GPQA Diamond leaderboard (1), HuggingFace MMLU-Pro leaderboard (1). No official leaderboard supplied any of them, so treat them as third-party measurements.
Models ranked nearby
- #453 gpt-5.6-luna-xhigh (codex-harness) (47.2)
- #454 seed-2.1-pro-preview (47.2)
- #455 MAmmoTH2-8x7B-Plus (47.2)
- #457 Qwen 3.6 Flash (2026-04-16) (47.2)
- #458 GPT-4o (47.2)
- #459 openhands-lm-32b-v0.1 (47.2)
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