BlogLLM Benchmark Scores as Published, Side by Side
Qwen3.8-27B: Benchmark Scores as Published
Every result the catalog holds for Qwen3.8-27B: 35 tests in 9 areas, from 4 pages. All were printed by Alibaba itself. Each number is as printed; nothing is ranked or rescored.
Developer AlibabaContext window 262KMax output 262KInput Text, image, videoOutput TextWeights Open (Apache 2.0)Data version 2026103001, as of 2026-10-30
Knowledge Work
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| JobBench | 33.4 | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| CoWorkBench | 70.7 | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
Agentic Coding
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| SWE-Bench Pro | 61.7% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | Harness: Claude Code |
| Terminal-Bench 2.1 | 73% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15 | Harness: Terminus |
| DeepSWE v1.1 | 42.2% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | Harness: Claude Code |
| SWE-bench Multilingual | 73.8% | Alibaba, Qwen3.8-Flash-Next model cardchecked 2026-10-15 | Harness: mini-SWE-agent; Context: 256000 |
| NL2Repo-Bench | 42.3 | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | Harness: Claude Code |
| QwenSWEBench | 79 | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15 | Harness: Claude Code |
| SWE-MM | 38.6% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15 | Harness: Claude Code |
Science and Math
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPQA Diamond | 89.2% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
Computer Use
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| OSWorld-Verified | 84.3% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15 | - |
| OSWorld 2.0 partial | 48% | Alibaba, Qwen3.8-Flash-Next model cardchecked 2026-10-15 | - |
| AndroidWorld | 81.9% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| OSWorld 2.0 strict | 19.4% | Alibaba, Qwen3.8-Flash-Next model cardchecked 2026-10-15 | - |
| RecreationBench | 47.1% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| WebArena-Verified | 64.8% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15 | - |
Multimodal Understanding
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| CharXiv Reasoning (no tools) | 83.7% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | Tools: none |
| MathVision | 90% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | Tools: none |
| CharXiv Reasoning (with tools) | 90.2% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | Tools: enabled; code interpreter |
| BabyVision (with tools) | 85.6% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15 | Tools: enabled; code interpreter |
| LVBench | 72.4% | Alibaba, Qwen3.8-Flash-Next model cardchecked 2026-10-15 | - |
| OmniDocBench 1.5 | 91.1% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15 | - |
| BabyVision (no tools) | 65.7% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15 | Tools: none |
| ERQA | 65.5% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| RealWorldQA | 85.9% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| MathVision (with tools) | 94.6% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | Tools: enabled; code interpreter |
General Knowledge and Reasoning
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Humanity's Last Exam (no tools) | 30.8% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
Code Generation
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| LiveCodeBench v6 | 90.3% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| Vision2Web | 62.9% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | Harness: Claude Code |
Tool Use and Workflows
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Agents' Last Exam (pass rate) | 20.4% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| Toolathlon Verified | 67.1% | Alibaba, Qwen3.8-Flash-Next model cardchecked 2026-10-15 | - |
| Agents' Last Exam (score) | 42.9% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| ClawEval-MM average score | 56.9 | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
| ClawEval-MM pass@3 | 57.4% | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
Instruction Following
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| IFBench | 79.5 | Alibaba, Qwen3.8-27B model cardchecked 2026-10-15; printed on 1 more page | - |
Price
API list prices in US dollars per one million tokens, as printed on the developer's pricing page on the day checked. Cost per task is measured by Vals AI.
| Price | Value | Printed by | Setup as printed |
|---|---|---|---|
| Input | $0.424 | Alibaba, Alibaba Model Studio pricingchecked 2026-10-15 | - |
| Output | $1.696 | Alibaba, Alibaba Model Studio pricingchecked 2026-10-15 | - |
| Cost per taskVals AI, USD, lower is better | $23.46 | Vals AI, Vibe Code Bench v1.1 leaderboardchecked 2026-10-27 | Harness: OpenHands |
More from Alibaba
- Qwen3.8-Max
- Qwen3.7-Plus
- Qwen3.6-27B
- Qwen3.6-35B-A3B
- Qwen3.6-Plus
- Qwen3.5-122B-A10B
- Qwen3.5-27B
- Qwen3.5-35B-A3B
- Qwen3.5-397B-A17B
- Qwen3-VL-32B-Instruct
- Qwen3-VL-32B-Thinking
- Qwen3-VL-8B-Instruct
- Qwen3-VL-8B-Thinking
- Qwen3-VL-235B-A22B
- Qwen3-VL-235B-A22B-Thinking
- Qwen3-Next-80B-A3B
- Qwen3-30B-A3B-Instruct-2507
- Qwen3-30B-A3B-Thinking-2507
- Qwen3-Coder-Plus
- Qwen3-235B-A22B
- Qwen3-32B
- Qwen3-8B
- QwQ-32B-Preview
- Qwen-Max
- Qwen3-Max-Thinking
- Qwen3.7-Max
- Qwen3.8-2.4T-A95B
- Qwen3.8-Flash-Next
Every number here is read from the same catalog file as the benchmark table, data version 2026103001. Where one page printed the model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.