- POPE 87.4
- MME-P 1567.1
- MMB 79.2
- SEED 72.1
- GQA 62
- MMMU 41
- MM-Vet 50
- GenEval Overall 0.80 80
- DPG-Bench Overall 84.2
Janus-Pro-7B
Janus-Pro-7B places in the top three on POPE.
20
Benchmarks with a score
0
Ranked first, out of five or more models
1
In the top three
DeepSeek Jan 2025 Janus-Pro-7B
Open weights Small
No Epoch Capabilities Index score yet
- MM-Vet 45.1% #100 /164
- SEED-Bench-2-Plus 53.9% #97 /136
- CoreCognition 43.9% #108 /119
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| POPE | 87.4% | #3 of 7 | Lab model card | |
| MMBench-CN v1.1 | 58.3% | #147 of 178 | Official leaderboard | |
| SimpleVQA | 31.3% | #11 of 13 | Official leaderboard |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GQA | 62% | #1 of 2 | Lab model card | |
| MM-Vet | 45.1% | #100 of 164 | Official leaderboard | |
| SEED-Bench | 72.1% | #2 of 3 | Lab model card | |
| CCBench (MMBench Chinese Culture) | 33.3% | #104 of 152 | Official leaderboard | |
| MMStar | 46.5% | #120 of 173 | Official leaderboard | |
| SEED-Bench-2-Plus | 53.9% | #97 of 136 | Official leaderboard | |
| A-Bench | 68.4% | #83 of 116 | Official leaderboard | |
| MMBench | 62.6% | #138 of 155 | Official leaderboard | |
| MMBench-EN v1.1 | 60% | #161 of 178 | Official leaderboard | |
| CoreCognition | 43.9% | #108 of 119 | Official leaderboard | |
| WISE | 0.33 | #10 of 11 | Official leaderboard | |
| WISE Verified | 0.33 | #10 of 11 | Official leaderboard | |
| MMMU | 41% | #101 of 103 | Lab model card | |
| MME | 1567.1 | #6 of 6 | Lab model card | |
| GenEval | 80% | #3 of 3 | Lab model card | |
| DPG-Bench | 84.2% | #1 of 1 | Lab model card |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OpenVLM Leaderboard | 50.2 | #124 of 173 | Official leaderboard |
Its lab's numbers next to 2 leaderboards
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.
Its system card
Every result in the longer documents the lab published with it.
Capabilities
Hallucination
- POPE 87.4