A state-of-the-art 12B small model built in collaboration with NVIDIA.
The most powerful model in its size category
Available under Apache 2.0 license
Multi-lingual (incl. European languages, Chinese, Japanese, Korean, Hindi, Arabic)
Large context window of 128K tokens
Top-tier reasoning for high-complexity tasks, for your most sophisticated needs.
Multi-lingual (incl. European languages, Chinese, Japanese, Korean, Hindi, Arabic)
Large context window of 128K tokens
Native function calling capacities and JSON outputs
High coding proficiency (80+ coding languages)
State-of-the-art Mistral model trained specifically for code tasks.
Trained on 80+ programming languages (incl. Python, Java, C++)
Optimized for low latency: smaller than coding models
Context window of 32K tokens
State-of-the-art semantic for extracting representation of text extracts.
English only for now
Achieves a retrieval score of 55.26 on the
Massive Text Embedding Benchmark (MTEB)
A Mamba2 language model designed for coding tasks.
Context window of 256K tokens
7.3B parameters
Available under Apache 2.0 license
Variant of Mistral-7B, optimized for solving advanced problems.
Context window of 32K tokens
7B parameters
Available under Apache 2.0 license
State-of-the-art semantic for extracting representation of text extracts.
Available in 8x7 and 8×22
sizes
Available under Apache 2.0 license
Access our latest products via our developer platform, hosted in Europe
La Plateforme is developers’ preferred way to access all Mistral Al’s models. Hosted and served on Mistral Al infrastructure, in Europe.
from ultrasafeai.client import ultrasafeClient
from ultrasafeai.models.chat_completion import ChatMessage
api_key = os.environ[“ULTRASAFE_API_KEY”]
model = “ultrasafe-tiny”
client = UltrasafeClient(api_key=api_key)
messages = [
ChatMessage(role=”user”,
content=”Who is the most renowned French painter?”)
]
We allow you to fine-tune our models in an easy, effective & cost-efficient way, and thus use smaller and better-suited models to solve your specific use cases. Fine-tuning can be done with our open-source fine-tuning code as well as on La Plateforme with our efficient Fine-tuning API.
Benefit from Ultrasafe fine-tuning code to perform fine-tuning on Ultrasafe open-source models on your own.
Leverage Ultrasafe’s unique expertise in training models by using our highly efficient fine-tuning service to specialize both our open-source and commercial models.
Get started with Mistral models in a few clicks via our developer platform hosted on Mistral’s infrastructure and build your own applications and services. Our servers are hosted in EU.
Access our models via your preferred cloud provider and use your cloud credits. Our open models are currently available via our cloud partners (GCP, AWS, Azure, Snowflake, NVIDIA). Mistral Large 2 is available on Azure AI Studio, AWS Bedrock, Google Cloud Model Garden and Snowflake.
Deploy Mistral models on virtual cloud or on-prem. Self-deployment offers more advanced levels of customisation and control. Your data stays within your walls. Try deploying our open models, and contact our team to deploy our optimized models similarly.
We distribute two categories of models:
| Apache 2.0 | Ultrasafe Research License | Ultrasafe Commercial License | |
|---|---|---|---|
| Access to weights | ✅ | ✅ | ✅ |
| Deployment for research purposes and individual usage | ✅ | ✅ | ✅ |
| Creation of derivatives (e.g. fine-tuning) for research purposes and individual usage | ✅ | ✅ The same license applies to derivatives | ✅ The same license applies to derivatives |
| Deployment for commercial purposes (internal & external use cases) | ✅ | ❌ Requires Ultrasafe Commercial License | ✅ |
| Creation and usage of derivatives (e.g. fine-tuning) for commercial use cases | ✅ | ❌ Requires Ultrasafe Commercial License | ✅ |
| Custom terms & support (self-deployment) | ❌ | ❌ | ✅ |
| Model | API Name | Description | Input | Output |
|---|---|---|---|---|
| Mistral Nemo | open-mistral-nemo-2407 | Mistral Nemo is a state-of-the-art 12B model developed with NVIDIA. | $0.3 /1M tokens | $0.3 /1M tokens |
| Mistral Large 2 | mistral-large-2407 | Top-tier reasoning for high-complexity tasks, for your most sophisticated needs. | $3 /1M tokens | $9 /1M tokens |
| Model | API Name | Description | Input | Output |
|---|---|---|---|---|
| Codestral | codestral-2405 | State-of-the-art Mistral model trained specifically for code tasks. | $1 /1M tokens | $3 /1M tokens |
| Mistral Embed | mistral-embed | State-of-the-art semantic for extracting representation of text extracts. | $0.1 /1M tokens |
| Model | One-off training | Storage | Input | Output |
|---|---|---|---|---|
| Mistral Nemo | $1 /1M tokens | $2 per month per model | $0.3 /1M tokens | $0.3 /1M tokens |
| Codestral | $3 /1M tokens | $2 per month per model | $1 /1M tokens | $3 /1M tokens |
| Mistral Large 2 | $9 /1M tokens | $4 per month per model | $3 /1M tokens | $9 /1M tokens |
| Model | API Name | Description | Input | Output |
|---|---|---|---|---|
| Mistral 7B | open-mistral-7b | A 7B transformer model, fast-deployed and easily customisable. | $0.25 /1M tokens | $0.25 /1M tokens |
| Mixtral 8x7B | open-mixtral-8x7b | A 7B sparse Mixture-of-Experts (SMoE). Uses 12.9B active parameters out of 45B total. | $0.7 /1M tokens | $0.7 /1M tokens |
| Mixtral 8x22B | open-mixtral-8x22b | Mixtral 8x22B is currently the most performant open model. A 22B sparse Mixture-of-Experts (SMoE). Uses only 39B active parameters out of 141B. | $2 /1M tokens | $6 /1M tokens |
| Mistral Small | mistral-small-latest | Cost-efficient reasoning, optimised for high volume use cases that require low latency. | $1 /1M tokens | $3 /1M tokens |
| Mistral Medium | mistral-medium-latest | Our first commercial model. | $2.75 /1M tokens | $8.1 /1M tokens |
Mistral AI provides a fine-tuning API through La Plateforme, making it easy to fine-tune our open-source and commercial models. There are three costs related to fine-tuning:
One-off training: Price per token on the data you want to fine-tune our standard models on; minimum fee per fine-tuning job of $4
Inference: Price per input/output token when using the fine-tuned model(s)
Storage: Price per month per model for storage (irrespective of model usage; models can be deleted any time)
