Cohere · AI model
Command A Vision
Cohere's first image-input Command model (July 2025) for chart and diagram analysis, table extraction, OCR and document Q&A; 128K context, 8K output, up to 20 images per request; tool use not supported.
Last verified Available 5 sources
Key specs
- Provider
- Cohere
- Family
- Command A
- API identifier
- command-a-vision-07-2025
- Availability
- Available
- Release date
- 2025-07-31
- License
- Open weights (CC-BY-NC 4.0 License with Acceptable Use Addendum)
- Context window
- 128,000 tokens
- Max output
- 8,000 tokens
- Knowledge cutoff
- June 2024
- API access
- Yes
- Consumer app
- No
Capabilities
Inputs
- Text input Yes
- Image input Yes Up to 20 images or 20MB per request (image inputs guide)
- Audio input No Docs model table lists modality as Text, Images
- Video input No Docs model table lists modality as Text, Images
- File input Unknown
Outputs
- Text output Yes
- Image output No Accepts images as input but does not generate them
- Audio output No Docs model table lists modality as Text, Images
Features
- Reasoning None Reasoning marked as not supported in the capabilities panel on the model's docs page
- Tool calling No Docs state tool use is not supported with this model
- Web search Unknown
"Unknown" means we have not verified this capability yet; it does not mean the model lacks it.
Pricing
No current prices listed.
Benchmarks
| Benchmark | Setting | Score | Reported by | Measured | Verified | Source |
|---|---|---|---|---|---|---|
| CharXiv Reasoning accuracy %, higher is better | CharXiv reasoning questions; tools, split and thinking mode not stated ('standard methodology') | 46.9 % | Self-reported | — | 2026-10-25 | Introducing Command A+ | Cohere |
There is no maintained public leaderboard (the project site says it is under construction), so almost all current numbers are self-reported by providers. Results with tools such as a code interpreter are much higher than without: Alibaba reports 83.7% without and 90.2% with tools for Qwen3.8-27B. Top models now exceed the human baseline, so differences at the top are small and may reflect label noise. Descriptive-question scores are a different metric and must not be recorded here.
One row per result, without rank numbers: each result was measured under its own setting (effort level, harness, tools, test version), as stated by the source, and scores compare only within the same setting. Self-reported results come from the provider and have not been reproduced independently.
Compare Command A Vision
Guides
About Command A Vision
Command A Vision was released on July 31, 2025 as Cohere's first model that accepts image input alongside text. It is aimed at chart, graph and diagram analysis, in-image table extraction, OCR and document question answering, and accepts up to 20 images or 20MB per request. It officially supports English, Portuguese, Italian, French, German and Spanish, and does not support tool use or image generation. On the Cohere API it is free up to the rate limits; production use requires contacting sales.
Strengths and weaknesses
Strengths
- Accepts text and image input (up to 20 images or 20MB per request; PNG, JPEG, WEBP and non-animated GIF)
- Documented for charts, diagrams, tables, OCR and document Q&A
- 128,000-token context window
Weaknesses
- Tool use is not supported
- Officially supports six languages
- No per-token API price is published; production use requires contacting sales
Sources
Last verified Page updated 2026-10-13