Mistral Large 3
Mistral Large 3 is a December 2025 open-weight mixture-of-experts model with 675B total and 41B active parameters, released under Apache 2.0. It accepts text and images and offers a 256K context window. Mistral positions it as a general-purpose model rather than a dedicated reasoning model.
Specifications
- Context window
- 256Kabout 197,000 words, or 390 pages
- Max output
- Not published
- Input price
- $0.50per 1M tokens
- Output price
- $1.50per 1M tokens
- Cached input
- Not published
- Weights
- OpenApache-2.0
- Parameters
- 675B total, 41B activeMoE
- Input types
- Text, Images
- Output types
- Text
- Reasoning mode
- No
- Tool calling
- Yes
- Released
- Dec 2, 2025Current
Mistral API price (mistral-large-2512) from Mistral docs; no cached-input price listed.
What it costs in practice
| Workload | Monthly cost at list price |
|---|---|
| Support chatbot30,000 conversations a month, 2,000 input and 400 output tokens each | $48 |
| Document summaries2,000 documents a month, 12,000 input and 800 output tokens each | $14 |
| Coding agent1,000 tasks a month, 60,000 input and 6,000 output tokens each | $39 |
Standard rates, no caching or batch discounts. Run your own numbers.
Strengths
- Apache 2.0 license at frontier scale
- Low first-party API price
- NVFP4 build runs on a single H100 or A100 node
Trade-offs to weigh
- Not a dedicated reasoning model
- FP8 needs a full B200 or H200 node to self-host
Where it fits
Mistral AI positions Mistral Large 3 for:
Where you can use it
- Hugging Face
- Ollama
- API
Run it yourself
Released under Apache-2.0. Approximate memory needed just to load the weights:
| Precision | Memory for weights |
|---|---|
| 16-bit (BF16/FP16) | 1,350 GB |
| 8-bit (FP8/INT8) | 675 GB |
| 4-bit (Q4_K_M, about 4.85 bits) | 409 GB |
This is a mixture-of-experts model: all 675B total parameters must sit in memory, but only about 41B are used per token, so it runs faster than a dense model of the same size.
Add memory for the KV cache (it grows with context length and concurrent users) and 1-2 GB of runtime overhead. Size it precisely in the VRAM calculator.
Alternatives to consider
| Model | Tier | Context | Input / output per 1M | |
|---|---|---|---|---|
| NVIDIA Nemotron 3 UltraNVIDIA | Advanced | 1M | $0.50 / $2.20 | Compare |
| DeepSeek-V4.1-FlashDeepSeek | Advanced | 1M | $0.30 / $1.20 | Compare |
| MiniMax-M3MiniMax | Advanced | 1M | $0.30 / $1.20 | Compare |
| gpt-oss-120bOpenAI | Advanced | 128K | $0.15 / $0.60 | Compare |
Frequently asked questions
How much does Mistral Large 3 cost?
Mistral Large 3 lists at $0.50 per million input tokens and $1.50 per million output tokens (checked October 10, 2026). Mistral API price (mistral-large-2512) from Mistral docs; no cached-input price listed. Confirm current rates on the provider's pricing page before you commit.
What is the context window of Mistral Large 3?
Mistral Large 3 accepts up to 262,144 tokens of context (about 197,000 words, or 390 pages). Long prompts cost more and can be slower, so most apps send only what the task needs.
Is Mistral Large 3 open source?
Mistral Large 3 is an open-weight model released under the Apache-2.0 license. You can download the weights from Hugging Face (mistralai/Mistral-Large-3-675B-Instruct-2512) and run or fine-tune them yourself, within the license terms. Open weights isn't the same as fully open source: training data and code may not be published.
Can I run Mistral Large 3 locally?
Yes, if you have enough memory. The weights alone need about 409 GB at 4-bit quantization and about 1,350 GB at 16-bit, plus room for the KV cache. Use the VRAM calculator on LLModel.com to check a specific GPU or Mac.
What is Mistral Large 3 best for?
Mistral AI positions Mistral Large 3 for enterprise, multilingual, self-hosting, agents. It accepts text, images as input, doesn't offer a separate reasoning mode and supports tool calling. Test it on a sample of your own tasks before you choose it for production.
Sources
- docs.mistral.ai/models/mistral-large-3-25-12
- huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512
- huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512/raw/main/params.json
- ollama.com/library
Facts checked against these pages on October 10, 2026. Spot an error? Tell us.
Still choosing a model?
Send us your use case and constraints. We'll come back with a shortlist and an evaluation plan you can run on your own data.
- Free and vendor-neutral: no model provider pays for placement.
- Your details go only to partners you agree to.