Fireworks RFT now available! Fine-tune open models that outperform frontier models. Try today

Model Library
/Mistral/Mistral 7B Instruct v0.3
Mistral Logo Icon

Mistral 7B Instruct v0.3

Ready
fireworks/mistral-7b-instruct-v3

    Mistral 7B Instruct v0.3 is an instruction fine-tuned version of the Mistral 7B v0.3 language model. It features an extended vocabulary of 32,768 tokens, supports the v3 tokenizer, and includes function calling capabilities. Optimized for instruction-following tasks, it excels in generating responses for chat applications, code generation, and understanding structured prompts. The model is designed to work efficiently with the Mistral inference library, enabling developers to deploy it across various natural language processing applications.

    Mistral 7B Instruct v0.3 API Features

    Fine-tuning

    Docs

    Mistral 7B Instruct v0.3 can be customized with your data to improve responses. Fireworks uses LoRA to efficiently train and deploy your personalized model

    On-demand Deployment

    Docs

    On-demand deployments give you dedicated GPUs for Mistral 7B Instruct v0.3 using Fireworks' reliable, high-performance system with no rate limits.

    Mistral 7B Instruct v0.3 FAQs

    What is Mistral 7B Instruct v0.3 and who developed it?

    Mistral 7B Instruct v0.3 is an instruction-tuned version of the Mistral 7B v0.3 base model, developed by Mistral AI. It supports the v3 tokenizer, includes function calling capabilities, and is optimized for instruction-following tasks like chat, code generation, and structured prompt handling.

    What applications and use cases does Mistral 7B Instruct v0.3 excel at?

    This model is designed for:

    • Conversational AI
    • Code assistance
    • Agentic systems
    • Structured prompt generation
    • Enterprise RAG and search
    What is the maximum context length for Mistral 7B Instruct v0.3?

    The model supports a context length of 32,768 tokens.

    What is the usable context window for Mistral 7B Instruct v0.3?

    The full 32.8K token window is supported on Fireworks on-demand deployments.

    What is the default temperature of Mistral 7B Instruct v0.3 on Fireworks AI?

    Users are expected to set decoding parameters manually (e.g., temperature, top_p) via API or CLI.

    What are known failure modes of Mistral 7B Instruct v0.3?
    • No moderation mechanisms: The model is not safety-aligned and may produce unfiltered or unsafe responses
    • No image input support, reranking, or embeddings
    • Requires updated tooling (transformers v4.42.0+) for function calling compatibility
    • Tool calling must include 9-character IDs for proper integration
    Does Mistral 7B Instruct v0.3 support streaming responses and function-calling schemas?
    • Streaming: Not supported
    • Function calling: Supported
    How many parameters does Mistral 7B Instruct v0.3 have?

    The model has 7.2 billion parameters.

    Is fine-tuning supported for Mistral 7B Instruct v0.3?

    Yes. Fireworks supports LoRA-based fine-tuning for this model on dedicated infrastructure.

    How are tokens counted (prompt vs completion)?

    Token usage is measured by the total of input and output tokens.

    What rate limits apply on the shared endpoint?
    • Serverless: Not supported
    • On-demand: Supported with no rate limits on dedicated GPUs
    What license governs commercial use of Mistral 7B Instruct v0.3?

    The model is released under the Apache 2.0 license, which permits unrestricted commercial use.

    Metadata

    State
    Ready
    Created on
    5/29/2024
    Kind
    Base model
    Provider
    Mistral
    Hugging Face
    Mistral-7B-Instruct-v0.3

    Specification

    Calibrated
    No
    Mixture-of-Experts
    No
    Parameters
    7.2B

    Supported Functionality

    Fine-tuning
    Supported
    Serverless
    Not supported
    Serverless LoRA
    Supported
    Context Length
    32.8k tokens
    Function Calling
    Supported
    Embeddings
    Not supported
    Rerankers
    Not supported
    Support image input
    Not supported