FlagEval EmbodiedVerse - CV-Bench (test): leaderboard
BAAI FlagEval’s EmbodiedVerse suite, CV-Bench (test) component: CV-Bench’s test split, classic 2D and 3D vision tasks recast as multiple choice. Scored on FlagEval’s common scale so the thirteen component boards read against each other.
Metric: Score. Source: flageval.baai.ac.cn. Saturation forecast: Around February 2027. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 235B A22B Instruct | 88.78 | #273 |
| 2 | Qwen 3 VL 235B A22B (Thinking) | 87.32 | #238 (Qwen 3 VL 235B A22B) |
| 3 | Qwen 3 VL 30B A3B Instruct | 86.78 | #384 |
| 4 | Qwen 3 VL 30B A3B (Thinking) | 86.7 | #351 (Qwen 3 VL 30B A3B) |
| 5 | GPT-5 | 85.77 | #98 |
| 6 | Gemini 2.5 Pro | 85.69 | #152 |
| 7 | InternVL3-78B | 82.74 | #352 |
| 8 | Qwen 2.5 VL 72B Instruct | 82.17 | #377 |
| 9 | GPT-4.1 | 82.01 | #248 |
| 10 | Qwen 2.5 VL 32B Instruct | 81.82 | #453 |
| 11 | InternVL3-8B | 80.14 | #620 |
| 12 | Claude Sonnet 4.5 | 80.08 | #142 |
| 13 | Qwen 2.5 VL 7B Instruct | 78.75 | #654 |
| 14 | Mistral Small 3.2 | 76.76 | #605 |
| 15 | O4 Mini | 64.9 | #180 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=flageval-embodiedverse-cv-bench-test · How It Works · Data refreshed daily, snapshot 2026-10-20.