Aya Vision 32B vs Magistral Medium vs Sonar Deep Research
Magistral Medium comes out ahead, 31 to 16 and 15 on our weighted score, and it is the cheaper option too.
CohereAya Vision 32B
15/100- ECI—
- Price—
- Context16K
- Our pick
Mistral AIMagistral Medium
31/100- ECI—
- Price$2.00 / $5.00
- Context128K
PerplexitySonar Deep Research
16/100- ECI—
- Price$2.00 / $8.00
- Context128K
Magistral Medium is our pick
Magistral Medium is the better all-round choice, scoring 31/100 against Sonar Deep Research (16) and Aya Vision 32B (15). It leads on inputs & features. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceMagistral MediumMagistral Medium $2.75 · Sonar Deep Research $3.50 per 1M tokens (3:1 blend) · Aya Vision 32B unpriced
- Longest contextMagistral Medium and Sonar Deep ResearchMagistral Medium 128,000 · Sonar Deep Research 128,000 · Aya Vision 32B 16,000 tokens
- Widest inputsAya Vision 32BAya Vision 32B: Text, Images · Magistral Medium: Text · Sonar Deep Research: Text
- Self-hostingAya Vision 32BPublishes downloadable weights (CC-BY-NC-4.0)
| Measure | Weight | Aya Vision 32B | Magistral Medium | Sonar Deep Research |
|---|---|---|---|---|
| Inputs & features | 60% | 25 | 35 | 10 |
| Context window | 40% | 0 | 24 | 24 |
| Overall | 100% | 15/100 | 31/100 | 16/100 |
Left out because at least one model lacks the data: capability and price. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Aya Vision 32BCohere | Magistral MediumMistral AI | Sonar Deep ResearchPerplexity |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | — | $2.00 | $2.00 |
| Output | — | $5.00 (best) | $8.00 |
| Cached input | — | — | — |
| Blended (3:1) | — | $2.75 (best) | $3.50 |
| Long-context rate | — | Same rate | Same rate |
| Price source | — | Official Mistral API | Official Perplexity API |
| Limits | |||
| Context window | 16,000 tokens | 128,000 tokens (best) | 128,000 tokens (best) |
| Max output | 4,000 tokens | 16,384 tokens | 32,768 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | No | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | Yes | Yesminimal · low · medium · high |
| Tool calling | No | Yes | No |
| Structured output | No | No | No |
| Availability | |||
| Weights | OpenCC-BY-NC-4.0 | Proprietary | Proprietary |
| API model ID | c4ai-aya-vision-32b | magistral-medium-latest | sonar-deep-research |
| API providers | 1 | 4 (best) | 3 |
| Released | Mar 4, 2025 | Mar 17, 2025 | Feb 1, 2025 |
| Knowledge cutoff | — | Jun 2025 | Jan 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Aya Vision 32B—
- Magistral Medium$30.00
- Sonar Deep Research$36.00
Which should you choose?
Which is better: Aya Vision 32B, Magistral Medium or Sonar Deep Research?
Magistral Medium is the better all-round choice, scoring 31/100 against Sonar Deep Research (16) and Aya Vision 32B (15). It leads on inputs & features. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Aya Vision 32B, Magistral Medium or Sonar Deep Research?
Magistral Medium is cheaper at $2.00 input / $5.00 output per million tokens (official Mistral API price). Sonar Deep Research costs $2.00 input / $8.00 output per million tokens (official Perplexity API price). At a typical mix of three input tokens to one output token, that is $2.75 per million tokens for Magistral Medium versus $3.50 for Sonar Deep Research (1.3× as much). Aya Vision 32B has no published per-token price.
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Aya Vision 32B has not been scored yet, Magistral Medium has not been scored yet and Sonar Deep Research has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Aya Vision 32B, Magistral Medium and Sonar Deep Research yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that Aya Vision 32B and Sonar Deep Research does not support tool calling, which most coding agents need.
Which has the bigger context window?
Magistral Medium and Sonar Deep Research have the largest context windows (128,000 and 128,000 tokens), against 16,000 for Aya Vision 32B. Maximum output per response: Aya Vision 32B up to 4,000, Magistral Medium up to 16,384, Sonar Deep Research up to 32,768 tokens.
Which can read images, PDFs, audio or video?
Aya Vision 32B accepts text and images; Magistral Medium accepts text; Sonar Deep Research accepts text. Aya Vision 32B handles the widest range of inputs.
Are any of these open source?
Aya Vision 32B publishes its weights (CC-BY-NC-4.0) and can be self-hosted; Magistral Medium and Sonar Deep Research is proprietary.
Which is newer?
Magistral Medium is the newest, released Mar 17, 2025. Aya Vision 32B came out Mar 4, 2025; Sonar Deep Research came out Feb 1, 2025. Knowledge cutoff: Magistral Medium Jun 2025, Sonar Deep Research Jan 2025.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.
