Qwen3.5 27B: benchmark scores, price and ranking
Qwen3.5 27B from Alibaba ranks #277 of 743 LLMs on the LLMs Tiger aggregate benchmark score (49.3), based on 8 benchmark results captured up to 2026-10-13.
- Organisation
- Alibaba
- Type
- Open-weight / local model
- Open weights
- Yes
- Released
- 2026-02-24 (source: epoch)
- Cheapest API price
- $0.260 in / $2.60 out per million tokens (via deepinfra, observed 2026-10-13T19:47Z)
- Aggregate rank
- #277 of 743 (score 49.3)
Relative to other models, Qwen3.5 27B is strongest on MMLU-Pro (#11 of 136) and weakest on Vending-Bench 2 (#16 of 19).
Across the 8 benchmarks where at least ten models are scored, Qwen3.5 27B is in the top 10% on 1 and in the top half on 4.
Its cheapest listed input price, $0.260 per million tokens, is lower than that of 63% of the 427 models with a listed price.
Qwen3.5 27B benchmark results
| Benchmark | Score | Rank | Source | Captured |
|---|---|---|---|---|
| MMLU-Pro | 86.1% | #11 of 136 | HuggingFace MMLU-Pro leaderboard | 2026-10-13 |
| GPQA Diamond | 85.5% | #77 of 275 | HuggingFace GPQA Diamond leaderboard | 2026-10-13 |
| DTBench | 82.4% | #44 of 146 | Epoch AI Benchmarking Hub | 2026-10-13 |
| LMCA | 34.0% | #46 of 108 | Epoch AI Benchmarking Hub | 2026-10-13 |
| WeirdML | 39.5% | #62 of 109 | Epoch AI Benchmarking Hub | 2026-10-13 |
| WebDev Arena | 1,357 | #45 of 67 | Epoch AI Benchmarking Hub | 2026-10-13 |
| ALE-Bench | 349 | #66 of 79 | Epoch AI Benchmarking Hub | 2026-10-13 |
| Vending-Bench 2 | $202 | #16 of 19 | Epoch AI Benchmarking Hub | 2026-10-13 |
About Qwen3.5 27B
Qwen3.5 27B is an open-weight model that can run on your own hardware from Alibaba. It was released on 2026-02-24, according to Epoch AI. The cheapest API listing we track charges $0.260 per million input tokens and $2.60 per million output tokens, through deepinfra. We hold 8 benchmark results for it across 4 areas, and 1 is in the top tenth of its board.
Where Qwen3.5 27B stands
On our aggregate score, Qwen3.5 27B is #277 of 743 ranked models, around the middle of the table. Its score is 49.3; the scale is centred on 50, so it sits 0.7 points below the typical model. That beats 63% of the ranked field. Among the 74 ranked Alibaba models it is #25. The next one up is Qwen3-Next-80B-A3B-Instruct at 49.3. The next one down is Qwen3-235B-A22B-Instruct-2507 at 49.3.
Qwen3.5 27B, benchmark by benchmark
On MMLU-Pro, Qwen3.5 27B scores 86.1% and ranks #11 of 136. Only 10 models score higher, led by Darwin-180B-RSI at 88.1%, which is 2.0 points ahead. It covers 57-subject knowledge + reasoning. The board median is 55.3%, so Qwen3.5 27B is 30.8 points above it. Source: HuggingFace MMLU-Pro leaderboard, captured 2026-10-13.
On GPQA Diamond, Qwen3.5 27B reaches 85.5%. It ranks #77 of 275, ahead of 72% of the field, and 10.3 points behind GPT-6 Astra (max) at 95.8%. The results either side are A.X-K2 (85.6%) and nemotron-3-ultra (85.4%). It covers expert-level science questions. The board median is 71.7%, so Qwen3.5 27B is 13.8 points above it. Source: HuggingFace GPQA Diamond leaderboard, captured 2026-10-13.
Qwen3.5 27B manages 82.4% on DTBench, good for #44 of 146. Claude Opus 5.5 (max) leads at 98.9%, 16.5 points further up, and 70% of models sit below. The results either side are grok-4-fast (82.7%) and Qwen3-Max-Instruct (82.1%). It covers dynamic tool-use benchmark. The board median is 68.8%, so Qwen3.5 27B is 13.6 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
LMCA puts Qwen3.5 27B at 34.0%, #46 of 108. That beats 57% of the models tested, but 34.3 points short of Claude Opus 5.5 (max) (68.2%). The results either side are GPT-5 mini (high) (34.2%) and MiniMax-M3 (33.7%). It covers language-model capability assessment. The board median is 29.2%, so Qwen3.5 27B is 4.8 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Qwen3.5 27B lands below the middle on WeirdML: 39.5%, #62 of 109, with 56% of the field scoring higher. GPT-6 Astra (max) leads at 93.3%; the closest result above is gpt-5-chat at 39.8%. It covers unusual machine-learning tasks. The board median is 41.2%, so Qwen3.5 27B is 1.6 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Qwen3.5 27B lands below the middle on WebDev Arena: 1,357, #45 of 67, with 66% of the field scoring higher. Claude Opus 5.5 (max) leads at 1,820; the closest result above is Gemma 4 26B A4B at 1,361. It covers building web apps, crowd-voted (Elo). The board median is 1,433, so Qwen3.5 27B is 76 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
ALE-Bench is a weak spot. Qwen3.5 27B scores 349, #66 of 79, behind 82% 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 Qwen3.5 27B is 415 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Vending-Bench 2 is a weak spot. Qwen3.5 27B scores $202, #16 of 19, behind 79% 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 Qwen3.5 27B is 2,175 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.
Strengths and weaknesses
Software engineering: weak, ahead of 25% of the field on average across 2 benchmarks. Its best board here is WebDev Arena (#45 of 67) and its weakest is ALE-Bench (#66 of 79). Reasoning, maths and science: strong, ahead of 82% of the field on average across 2 benchmarks. Its best board here is MMLU-Pro (#11 of 136) and its weakest is GPQA Diamond (#77 of 275). General knowledge and chat: uneven, ahead of 39% of the field on average across 3 benchmarks. Its best board here is LMCA (#46 of 108) and its weakest is Vending-Bench 2 (#16 of 19).
Price and value
The median input price across the 427 priced models is $0.500 per million tokens, so Qwen3.5 27B 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), Qwen3.7 Flash ($0.030, 50.5). Similar scores cost more on Qwen3.7 Plus ($0.282) and gemini-exp-1206 ($0.300).
Release history
The Alibaba model released before it was Qwen3.5 Plus on 2026-02-16, which scores 49.6. The next one was qwen3.5-35b-a3b on 2026-02-24, at 48.2.
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
- #274 Qwen3-Next-80B-A3B-Instruct (49.3)
- #275 gemini-2.5-flash-preview-09-2025_16k (49.3)
- #276 Qwen3-4B-Thinking-2507 (49.3)
- #278 Qwen3-235B-A22B-Instruct-2507 (49.3)
- #279 qwq-plus (49.3)
- #280 Mistral-Large-Instruct-2407 (49.3)
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