Very long inputs
Choose Ministral 3 8B
Ministral 3 8B reads up to 256k tokens per request, against 8k for Tiny Aya Global.
Ministral 3 8B vs Tiny Aya Global: price, context and capabilities
Choose Ministral 3 8B for very long inputs and images as input. Choose Tiny Aya Global for the latest release.
In short
- A price comparison is not possible: one of them has no comparable published price.
- Ministral 3 8B reads more per request: 256k tokens against 8k.
- Tiny Aya Global is the newer model (Feb 17, 2026, against December 2025).
- Only Ministral 3 8B offers vision, tool calling.
Which should you choose?
Each answer below comes straight from the published prices, specifications and independent benchmark results.
Images as input
Choose Ministral 3 8B
Ministral 3 8B accepts images as input; Tiny Aya Global does not.
Latest release
Choose Tiny Aya Global
Tiny Aya Global is the newer release (Feb 17, 2026, against December 2025 for Ministral 3 8B).
Specifications side by side
| Ministral 3 8B | Tiny Aya Global | |
|---|---|---|
| Provider | Mistral AI | Cohere |
| Category | Language models | Language models |
| Status | Generally available | Generally available |
| Headline price | No data | No data |
| Input | No data | No data |
| Output | No data | No data |
| Cached input | No data | No data |
| Context window | 256k | 8k |
| Max output | No data | No data |
| Released | December 2025 | Feb 17, 2026 |
| Knowledge cutoff | No data | No data |
| Licence | Open weights | Open weights |
| Capabilities | Vision, Tool calling, Self-hostable | Self-hostable |
| API model id | ministral-8b-2512 | tiny-aya-global |
Benchmarks side by side
Leader in bold.
| Benchmark | Ministral 3 8B | Tiny Aya Global |
|---|---|---|
| GPQA Diamond | 47.1% | 30.5% |
| Humanity's Last Exam | 4.3% | 5.2% |
| IFBench (instruction following) | 29.1% | 20.1% |
| Intelligence Index | 5.5 | No data |
| Terminal-Bench Hard | 4.5% | No data |
| SciCode | 20.7% | No data |
| τ²-Bench (agentic tool use) | 26.6% | No data |
| MMMU-Pro (visual reasoning) | 46% | No data |
| Long-context reasoning (AA-LCR) | 25.7% | No data |
Independent evaluations by Artificial Analysis, run the same way for every model, at the reasoning setting shown. Percentages unless marked as an index.
Frequently asked questions
Should I choose Ministral 3 8B or Tiny Aya Global?
Choose Ministral 3 8B for very long inputs and images as input. Choose Tiny Aya Global for the latest release.
Which is cheaper, Ministral 3 8B or Tiny Aya Global?
A price comparison is not possible: one of them has no comparable published price.
Which has the larger context window, Ministral 3 8B or Tiny Aya Global?
Ministral 3 8B reads more per request: 256k tokens against 8k.
What is the difference between Ministral 3 8B and Tiny Aya Global?
Ministral 3 8B is made by Mistral AI and Tiny Aya Global by Cohere. A price comparison is not possible: one of them has no comparable published price. Ministral 3 8B reads more per request: 256k tokens against 8k. Tiny Aya Global is the newer model (Feb 17, 2026, against December 2025). Only Ministral 3 8B offers vision, tool calling.
Which is newer, Ministral 3 8B or Tiny Aya Global?
Tiny Aya Global was released on Feb 17, 2026 and Ministral 3 8B on December 2025.