Individual benchmark scores plotted by date.
| Organisation | Model | Reported | Top Score | Info | Self Reported | Source |
|---|---|---|---|---|---|---|
| Seed 2.0 Pro | 14 Feb 2026 | 79.50% | Seed2 official benchmark table | BLINK | Yes | Source | |
| Seed 2.0 Lite | 14 Feb 2026 | 75.60% | Seed2 official benchmark table | BLINK | Yes | Source | |
| Seed 2.0 Mini | 14 Feb 2026 | 73.40% | Seed2 official benchmark table | BLINK | Yes | Source | |
| Qwen 3 VL 235B A22B Instruct | - | 70.70% | - | Yes | - | |
| Qwen 3 VL 8B Instruct | - | 69.10% | - | Yes | - | |
| Qwen 3 Reranker 8B | - | 68.70% | inferred family alias from qwen3-vl-8b-thinking (score=0.3850; benches=50) | Yes | - | |
| Qwen 3 Embedding 8B | - | 68.70% | inferred family alias from qwen3-vl-8b-thinking (score=0.3850; benches=50) | Yes | - | |
| Qwen 3 8B | - | 68.70% | inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.4600; benches=50) | Yes | - | |
| Qwen 3 Guard Gen 8B | - | 68.70% | inferred family alias from qwen3-vl-8b-thinking (score=0.3400; benches=50) | Yes | - | |
| Qwen 3 VL Embedding 8B | - | 68.70% | inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.5232; benches=50) | Yes | - | |
| Qwen 3 VL 8B Thinking | - | 68.70% | - | Yes | - | |
| Qwen 3 VL Reranker 8B | - | 68.70% | inferred high-confidence family alias from qwen3-vl-8b-thinking (score=0.5275; benches=50) | Yes | - | |
| Qwen 3 Guard Stream 8B | - | 68.70% | inferred family alias from qwen3-vl-8b-thinking (score=0.3371; benches=50) | Yes | - | |
| Qwen 3 VL 32B Thinking | - | 68.50% | - | Yes | - | |
| Qwen 3 VL 30B A3B Instruct | - | 67.70% | - | Yes | - | |
| Qwen 3 VL 32B Instruct | - | 67.30% | - | Yes | - | |
| Qwen 3 VL 235B A22B Thinking | - | 67.10% | - | Yes | - | |
| Qwen 3 VL 4B Instruct | - | 65.80% | - | Yes | - | |
| Qwen 3 VL 30B A3B Thinking | - | 65.40% | - | Yes | - | |
| Qwen 3 Reranker 4B | - | 63.40% | inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48) | Yes | - | |
| Qwen 3 4B Thinking 2507 | - | 63.40% | inferred family alias from qwen3-vl-4b-thinking (score=0.3462; benches=48) | Yes | - | |
| Qwen 3 Guard Gen 4B | - | 63.40% | inferred family alias from qwen3-vl-4b-thinking (score=0.3400; benches=48) | Yes | - | |
| Qwen 3 4B Instruct 2507 | - | 63.40% | inferred family alias from qwen3-vl-4b-thinking (score=0.3462; benches=48) | Yes | - | |
| Qwen 3 Embedding 4B | - | 63.40% | inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48) | Yes | - | |
| Qwen 3 Guard Stream 4B | - | 63.40% | inferred family alias from qwen3-vl-4b-thinking (score=0.3371; benches=48) | Yes | - | |
| Qwen 3 VL 4B Thinking | - | 63.40% | - | Yes | - | |
| Qwen 3 4B SafeRL | - | 63.40% | inferred family alias from qwen3-vl-4b-thinking (score=0.3850; benches=48) | Yes | - | |
| Phi 4 multimodal instruct | 01 Feb 2025 | 61.30% | - | Yes | Source |
