Alibaba Cloud / Qwen Team model product
Qwen3 VL 8B Thinking is a multimodal language model for understanding text, images, and video with a focus on reasoning and STEM tasks.
Updated Sep 8, 2026. Default version: Qwen3 VL 8B Thinking
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Benchmark scores for Qwen3 VL 8B Thinking.
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkMuirBench | Score76.80% | Rank04 | Participants12 | Percentile72.73% | EvidenceC | Evaluated |
| BenchmarkBLINK | Score68.70% | Rank05 | Participants15 | Percentile71.43% | EvidenceC | Evaluated |
| BenchmarkMMMU (val) | Score74.10% | Rank05 | Participants13 | Percentile66.67% | EvidenceC | Evaluated |
| BenchmarkOCRBench-V2 (en) | Score63.90% | Rank06 | Participants14 | Percentile61.54% | EvidenceC | Evaluated |
| BenchmarkWritingBench | Score85.50% | Rank06 | Participants15 | Percentile64.29% | EvidenceC | Evaluated |
| BenchmarkCharadesSTA | Score59.90% | Rank07 | Participants12 | Percentile45.45% | EvidenceC | Evaluated |
| BenchmarkInfoVQAtest | Score86.00% | Rank07 | Participants12 | Percentile45.45% | EvidenceC | Evaluated |
| BenchmarkMLVU-M | Score75.10% | Rank07 | Participants8 | Percentile14.29% | EvidenceC | Evaluated |
| BenchmarkMM-MT-Bench | Score8.00 points | Rank07 | Participants17 | Percentile62.50% | EvidenceC | Evaluated |
| BenchmarkOCRBench-V2 (zh) | Score59.20% | Rank07 | Participants11 | Percentile40.00% | EvidenceC | Evaluated |
| BenchmarkDocVQAtest | Score95.30% | Rank08 | Participants11 | Percentile30.00% | EvidenceC | Evaluated |
| BenchmarkHallusion Bench | Score65.40% | Rank08 | Participants18 | Percentile58.82% | EvidenceC | Evaluated |
| BenchmarkScreenSpot | Score93.60% | Rank09 | Participants16 | Percentile46.67% | EvidenceC | Evaluated |
| BenchmarkCharXiv-D | Score85.90% | Rank10 | Participants17 | Percentile43.75% | EvidenceC | Evaluated |
| BenchmarkLiveBench 20241125 | Score69.80% | Rank10 | Participants14 | Percentile30.77% | EvidenceC | Evaluated |
| BenchmarkMMBench-V1.1 | Score87.50% | Rank10 | Participants20 | Percentile52.63% | EvidenceC | Evaluated |
| BenchmarkMMStar | Score75.30% | Rank11 | Participants25 | Percentile58.33% | EvidenceC | Evaluated |
| BenchmarkMulti-IF | Score75.10% | Rank11 | Participants23 | Percentile54.55% | EvidenceC | Evaluated |
| BenchmarkCreative Writing v3 | Score82.40% | Rank12 | Participants13 | Percentile8.33% | EvidenceC | Evaluated |
| BenchmarkMathVista-Mini | Score81.40% | Rank13 | Participants25 | Percentile50.00% | EvidenceC | Evaluated |
| BenchmarkOSWorld | Score33.90% | Rank13 | Participants20 | Percentile36.84% | EvidenceC | Evaluated |
| BenchmarkVideo-MME | Score71.80% | Rank14 | Participants17 | Percentile18.75% | EvidenceC | Evaluated |
| BenchmarkODinW | Score39.80% | Rank15 | Participants16 | Percentile6.67% | EvidenceC | Evaluated |
| BenchmarkPolyMATH | Score47.50% | Rank15 | Participants23 | Percentile36.36% | EvidenceC | Evaluated |
| BenchmarkSimpleQA | Score49.60% | Rank15 | Participants47 | Percentile69.57% | EvidenceC | Evaluated |
| BenchmarkArena-Hard v2 | Score51.10% | Rank16 | Participants19 | Percentile16.67% | EvidenceC | Evaluated |
| BenchmarkCC-OCR | Score76.30% | Rank16 | Participants18 | Percentile11.76% | EvidenceC | Evaluated |
| BenchmarkMVBench | Score69.00% | Rank16 | Participants18 | Percentile11.76% | EvidenceC | Evaluated |
| BenchmarkBFCL-v3 | Score63.00% | Rank19 | Participants19 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkOCRBench | Score81.90% | Rank20 | Participants24 | Percentile17.39% | EvidenceC | Evaluated |
Preference and agent-evaluation results for the default version.
The default version does not have a matching Arena result yet.
Official vendor API pricing appears first, followed by individual provider offers.
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderOpenRouter | Provider model IDqwen/qwen3-vl-8b-thinking | Regionglobal | Input / 1M$0.18 | Output / 1M$2.1 | Context131.1K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Provider-specific output speed and catalog latency for Qwen3 VL 8B Thinking. Runtime does not affect the capability score.
No provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Technical details for the model's default version.
Available versions of this model. The score column identifies the version used in the overall ranking.
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionQwen3 VL 8B Thinking | Released | LLMBoard27.25 | Parameters9B | Context262.1K | Max output262.1K | Open weightsYes | LicenseApache 2.0 |
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
Key information about Qwen3 VL 8B Thinking and its available data.
Qwen3 VL 8B Thinking is a model from Alibaba Cloud / Qwen Team that combines vision, language, and reasoning. It supports visual understanding, spatial reasoning, long-video comprehension, multilingual OCR, and tool-based interaction.
Data as of 2026-09-08.
Common questions about Qwen3 VL 8B Thinking.
Qwen3 VL 8B Thinking's default version was released on Sep 22, 2025.
No official standard PAYG price is currently available for Qwen3 VL 8B Thinking. The lowest tracked third-party offer starts at $0.18 input and $2.1 output via OpenRouter.
Qwen3 VL 8B Thinking was created by Alibaba Cloud / Qwen Team.
The default version has a 262.1K token context window.
Yes. The default version is marked as open weight under Apache 2.0.
1 provider offerings are linked to the default version.