- MMBench DEV_EN_V11 (VLMEvalKit) 68.7
- MMStar 51.8
- RealWorldQA 62.2
- GQA TestDev_Balanced 57.4
- MMMB 72.8
- BLINK 52.7
- ChartQA Test 80.8
- DocVQA VAL 92.1
- InfoVQA VAL 65.2
- OCRBench v2_en 36.7
- OCRBench (0.792) 79.2
- AI2D_TEST 77.5
- MMMU DEV_VAL 32.9
- RefCOCO avg (average of 8 RefCOCO, RefCOCO+ and RefCOCOg splits) 73.2
- CountBench 72.5
- HallusionBench 61.5
- MMLU test 50.4
- MMLU-Pro test 30.7
- Multi-If 37.3
- IFEval 74.9
North Micro Vision Instruct
North Micro Vision Instruct.
20
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
Cohere Aug 2026 North Micro Vision Instruct
Open weights Small
111.1
#375 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
- HallusionBench 61.5% #21 /31
- BLINK 52.7% #25 /29
- MMLU-Pro 30.7% #165 /167
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
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DocVQA | 92.1% | #24 of 36 | Lab model card | |
| InfographicVQA | 65.2% | #21 of 26 | Lab model card | |
| BLINK | 52.7% | #25 of 29 | Lab model card | |
| ChartQA | 80.8% | #33 of 37 | Lab model card | |
| CountBench | 72.5% | #25 of 28 | Lab model card | |
| OCRBench | 79.2% | #31 of 34 | Lab model card | |
| RealWorldQA | 62.2% | #35 of 38 | Lab model card | |
| AI2D | 77.5% | #49 of 52 | Lab model card | |
| MMStar | 51.8% | #35 of 36 | Lab model card | |
| MMMU | 32.9% | #103 of 103 | Lab model card | |
| RefCOCO / RefCOCO+ / RefCOCOg | 73.2% | #29 of 29 | Lab model card | |
| MMBench-EN v1.1 | 68.7% | #23 of 23 | Lab model card | |
| OCRBench v2 | 36.7% | #22 of 22 | Lab model card | |
| GQA | 57.4% | #2 of 2 | Lab model card |
Its lab's numbers put it at 111.1, number 375 of 404
The Self-Reported Index fits one score to the 18 capability results Cohere reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 107.1 to 115.0. How we compute it.
- DocVQA 92.1
- ChartQA 80.8
- OCRBench 79.2
- AI2D 77.5
- IFEval 74.9
- RefCOCO / RefCOCO+ / RefCOCOg 73.2
- CountBench 72.5
- MMBench-EN v1.1 68.7
- InfographicVQA 65.2
- RealWorldQA 62.2
- HallusionBench 61.5
- BLINK 52.7
- MMStar 51.8
- MMLU 50.4
- Multi-IF 37.3
- OCRBench v2 36.7
- MMMU 32.9
- MMLU-Pro 30.7
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.