Ministral 3 3B 2512
- Context
- 131K
- Released
- 10mo ago
- Parameters
- 3B
- Input / 1M tokens
- $0.10
- Output / 1M tokens
- $0.10
- Cached input / 1M tokens
- $0.01
Sources:OpenRouter
Benchmarks
DeepSWE is the primary coding benchmark where an exact model evaluation exists. Other sourced evaluations appear below; missing scores are not zero.
Not evaluated in DeepSWE 1.1 as of 2026-09-22. Results are only shown for exact model versions in the official dataset. Official leaderboard.
No benchmark results have been matched to this model yet.
Specifications
- Developer
- Mistral AI
- Model family
- Ministral
- Type
- Language model
- Predecessor
- —
- Successor
- —
- Released
- December 2, 2025
- Model ID
mistralai/ministral-3b-2512- Openness
- Open weights
- License
- —
- Parameters
- 3B
- Input modalities
- text, image
- Output modalities
- text
- Modality
- text+image->text
- Native context
- 256K
- Tokenizer
- Mistral
- Max output tokens
- 105K
- Hugging Face
- mistralai/Ministral-3-3B-Instruct-2512
- Source freshness
- Endpoints synced Oct 15, 2026, 12:43 PM UTC
Open-weight builds
Ollama variants
Every published Ollama tag matched to this model, including parameter editions, instruction/base variants, and quantized builds.
| Variant | Parameters | Type | Quantization | Size | Context | Input |
|---|---|---|---|---|---|---|
| ministral-3:3b-instruct-2512-fp16 | 3B | instruct | FP16 | 7.7 GB | 256K | text + image |
| ministral-3:3b | 3B | instruct | Q4_K_M | 3 GB | 256K | text + image |
| ministral-3:3b-instruct-2512-q8_0 | 3B | instruct | Q8_0 | 4.5 GB | 256K | text + image |
| ministral-3:8b-instruct-2512-fp16 | 8B | instruct | FP16 | 18 GB | 256K | text + image |
| ministral-3:latest | 8B | instruct | Q4_K_M | 6 GB | 256K | text + image |
| ministral-3:8b-instruct-2512-q8_0 | 8B | instruct | Q8_0 | 9.9 GB | 256K | text + image |
| ministral-3:14b-instruct-2512-fp16 | 14B | instruct | FP16 | 28 GB | 256K | text + image |
| ministral-3:14b | 14B | instruct | Q4_K_M | 9.1 GB | 256K | text + image |
| ministral-3:14b-instruct-2512-q8_0 | 14B | instruct | Q8_0 | 15 GB | 256K | text + image |
Infrastructure
Provider availability
| Provider | Endpoint | Context | Input /M tokens | Output /M tokens | Cached input /M tokens | Cache write /M tokens | Uptime 1d | Data handling |
|---|---|---|---|---|---|---|---|---|
| mistral/zdr | 131K | $0.10 | $0.10 | $0.01 | — | 100.00% | ZDR: YesTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| mistral | 131K | $0.10 | $0.10 | $0.01 | — | 100.00% | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| mistral/eu | 131K | $0.11 | $0.11 | $0.01 | — | 100.00% | ZDR: YesTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 |
Provider data:OpenRouter
Deployment graph
Can I run this?
Vertical Tabs only shows compatibility when it has structured evidence tying this exact model to hardware, runtime, quantization, memory use, and observed performance.