MMEB-V2
Qwen3-VL-Embedding-8B 77.9% leads since January 2026 Qwen3-VL-Embedding-8B 77.8% leads Qwen3-VL-Embedding-8B 77.9% leads since January 2026 Multimodal embedding models are scored on 78 tasks over images, videos and visual documents. V2 adds visual document retrieval, video retrieval, temporal grounding, video classification and video QA.
Released with VLM2Vec-V2, accepted to TMLR 2026. Google's EmbeddingGemma 2 card reports its Image, VisDoc and Video splits.
- Released
- May 2025
- Kind
- General visual questions
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- All public
- Last in a launch post
- January 2026, dropped before it saturated
- Last model added
- October 2026
- Scores refreshed
- November 12th, 2026
Qwen3-VL-Embedding-8B leads with 77.9%
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
Qwen3-VL-Embedding-8B leads with 77.8% across all independent runs
Best score per model across the official leaderboard, Epoch AI’s own runs and any board Epoch AI copied. Each bar and row names its source.
Qwen3-VL-Embedding-8B leads with 77.9%, claims included
Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model.
- 1 Qwen3-VL-Embedding-8B 77.9%
- 2 Qwen3-VL-Embedding-2B 73.2%
- 3
EmbeddingGemma 2 57.3%
Source: Labs’ model cards
- 1 Qwen3-VL-Embedding-8B 77.8%
- 2 Qwen3-VL-Embedding-2B 73.3%
Source: Official MMEB-V2 leaderboard
- 1 Qwen3-VL-Embedding-8B 77.9%
- 2 Qwen3-VL-Embedding-2B 73.3%
- 3 EmbeddingGemma 2 57.3%
Sources: Labs’ model cards and official MMEB-V2 leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Qwen3-VL-Embedding-8B | 77.9% | Jan 2026 | Lab model card | |
| 2 | Qwen3-VL-Embedding-2B | 73.2% | Jan 2026 | Lab model card | |
| 3 | EmbeddingGemma 2 | 57.3% | Oct 2026 | Lab model card |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Qwen3-VL-Embedding-8B | 77.8% | Jan 2026 | Official leaderboard | |
| 2 | Qwen3-VL-Embedding-2B | 73.3% | Jan 2026 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Qwen3-VL-Embedding-8B | 77.9% | Jan 2026 | Lab model card | |
| 2 | Qwen3-VL-Embedding-2B | 73.3% | Jan 2026 | Official leaderboard | |
| 3 | EmbeddingGemma 2 | 57.3% | Oct 2026 | Lab model card |
The benchmark's own leaderboard has 45 entries, 2 of them models we track. Switch the scores above to rank them together.
Number one over time Number one over time Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Labs’ model cards
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Sources: Official MMEB-V2 leaderboard and labs’ model cards
Labs headline it less often now, last in January 2026
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| Jan 8, 2026 | Qwen3-VL-Embedding-8B | 77.9 | MMEB-V2 All, image 80.1, video 66.1, VisDoc 83.3 (8B card table, VisDoc OOD split re-evaluated) |
| Jan 8, 2026 | Qwen3-VL-Embedding-2B | 73.2 | MMEB-V2 All, image 75.0, video 61.9, VisDoc 79.2 (2B card table) |
One system card reports it
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| Oct 6, 2026 | EmbeddingGemma 2 model card | 57.3 | MMEB v2 Image, Mean(Task), Hit@1, full precision, 768d | Capabilities |
2 lab claims next to the leaderboard
Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. 2 of them match, which usually means one side copied the other. Every benchmark.
| Model | Lab says | Board says | Gap |
|---|---|---|---|
| Qwen3-VL-Embedding-8B | 77.9% MMEB-V2 All, image 80.1, video 66.1, VisDoc 83.3 (8B card table, VisDoc OOD split re-evaluated) | 77.8% Leaderboard | Same |
| Qwen3-VL-Embedding-2B | 73.2% MMEB-V2 All, image 75.0, video 61.9, VisDoc 79.2 (2B card table) | 73.3% Leaderboard | Same |
Versions
More in Vision
- MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. GPT-5.1 85.4%
- MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
- CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
- MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
- MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%
- ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance.
Command A Vision 90.9%