Mistral Provider

Overview

The Mistral provider gives you access to Mistral AI’s models for chat completion and embeddings, optimized for European languages and multilingual tasks.

Installation

npm install @core-ai/mistral

createMistral()

Create a Mistral provider instance.

import { createMistral } from '@core-ai/mistral';

const mistral = createMistral({
    apiKey: process.env.MISTRAL_API_KEY,
});

Options

string

Your Mistral API key. Defaults to MISTRAL_API_KEY environment variable.

string

Custom base URL for API requests.

Mistral

Provide your own configured Mistral client instance.

Returns

MistralProvider with methods chatModel() and embeddingModel().

Supported models

Chat models

Large models

mistral-large-4 and mistral-large-4-0 — Mistral Large 4 (public preview). Multimodal, 1M context, adjustable reasoning
mistral-large-latest — latest general-availability Large model (Large 3, mistral-large-2512)

Medium models

mistral-medium-3-5, mistral-medium-3, and mistral-medium-latest — Mistral Medium 3.5. Multimodal, adjustable reasoning

Small models

mistral-small-2603 and mistral-small-latest — Mistral Small 4. Multimodal, adjustable reasoning

Other chat models

zai-glm-5-3, zai-glm-5, and zai-glm-latest — Z.ai GLM 5.3, hosted on the Mistral API. Text only, adjustable reasoning (low, high, max)
codestral-latest — code generation

Embedding models

mistral-embed - High-quality text embeddings

Provider support

Feature	Support
Chat Completion	Yes
Streaming	Yes
Function Calling	Yes
Vision	Yes
Reasoning	Adjustable on Small 4, Medium 3.5, Large 4, and GLM 5.3
Embeddings	Yes
Image Generation	No

Model capabilities

reasoning.mode is optional when the model accepts reasoning_effort, and unsupported otherwise. Omitting reasoning does not send the parameter.

Models	IDs	`supportedEfforts`	Sent as
Small 4	`mistral-small-latest`, `mistral-small-2603`, and later `mistral-small-YYMM`	`minimal`, `high`	`minimal` and `low` → `none`; `medium`, `high`, and `max` → `high`
Medium 3.5	`mistral-medium-latest`, `mistral-medium-3`, `mistral-medium-3-5`, and `mistral-medium-YYMM` from `2604`	`minimal`, `high`	same mapping as Small 4
Large 4	`mistral-large-4`, `mistral-large-4-0`	`minimal`, `high`	same mapping as Small 4
GLM 5.3	`zai-glm-5-3`, `zai-glm-5`, `zai-glm-latest`, `zai-glm-5-2`	`low`, `high`, `max`	`minimal` and `medium` → `low`; `low`, `high`, and `max` pass through

Every other model id reports reasoning.mode: 'unsupported'. reasoning.effort is ignored, and reasoning_effort is not sent. Thinking chunks in a response are still returned as reasoning parts. Inspect the model instance or use the exported helper:

import { getMistralModelCapabilities } from '@core-ai/mistral';

const model = mistral.chatModel('mistral-small-latest');
const capabilities = model.capabilities;

const otherCapabilities = getMistralModelCapabilities('mistral-large-4');

modalities.input is resolved per model. Vision families such as mistral-large, mistral-medium, mistral-small, magistral-*, ministral-*, and pixtral-* advertise ['text', 'image', 'file'] (image sources: base64 and url); text-only families such as codestral, devstral-*, open-mistral-*, open-mixtral-*, and zai-glm-* advertise ['text'] and reject images. Unrecognized model IDs are treated as multimodal capable. modalities.output is ['text'] for all chat models. Adjustable-reasoning ids are matched first. Everything else prefers an exact model ID and then falls back to the family, ignoring the -latest alias and the -YYMM version suffix. That family fallback is how vision is resolved: pinning an older version reports the capabilities of that version rather than of -latest:

Family	Vision from	Earlier versions reported text-only
Mistral Large	Large 3 (`-2512`)	`-2402`, `-2407`, `-2411`
Mistral Medium	Medium 3 (`-2505`)	`-2312`
Mistral Small	Small 3.1 (`-2503`)	`-2402`, `-2409`, `-2501`
Magistral	Magistral 1.2 (`-2509`)	`-2506`, `-2507`
Ministral	Ministral 3 (`-2512`)	`-2410`

Strict tool schemas

Mistral’s API accepts a strict flag but does not document enforcement, so every model reports model.capabilities.tools.strictSchemas as { supported: false } and strict: true throws ToolSchemaStrictnessError locally. See Strict tool schemas.

Examples

Basic chat

import { createMistral } from '@core-ai/mistral';
import { generate } from '@core-ai/core-ai';

const mistral = createMistral();

const result = await generate({
    model: mistral.chatModel('mistral-large-2512'),
    messages: [{ role: 'user', content: 'Explain the concept of recursion' }],
});

console.log(result.content);

Streaming

import { stream } from '@core-ai/core-ai';

const chatStream = await stream({
    model: mistral.chatModel('mistral-medium-2508'),
    messages: [
        {
            role: 'user',
            content: 'Write a short story about artificial intelligence',
        },
    ],
});

for await (const event of chatStream) {
    if (event.type === 'text-delta') {
        process.stdout.write(event.text);
    }
}

Reasoning output

const result = await generate({
    model: mistral.chatModel('mistral-small-latest'),
    messages: [
        {
            role: 'user',
            content: 'Solve this step by step: if 3x + 7 = 22, what is x?',
        },
    ],
    reasoning: {
        effort: 'high',
    },
});

console.log(result.reasoning);

effort: 'high' is sent as reasoning_effort: 'high'. effort: 'minimal' is sent as none, which omits the thinking trace. Native Mistral reasoning is replayed as a thinking chunk on the next turn.

Tool calling

import { generate, defineTool } from '@core-ai/core-ai';
import { z } from 'zod';

const result = await generate({
    model: mistral.chatModel('mistral-large-2512'),
    messages: [{ role: 'user', content: 'What is 25 multiplied by 4?' }],
    tools: {
        calculator: defineTool({
            name: 'calculator',
            description: 'Perform mathematical calculations',
            parameters: z.object({
                operation: z.enum(['add', 'subtract', 'multiply', 'divide']),
                a: z.number(),
                b: z.number(),
            }),
        }),
    },
});

Embeddings

import { embed } from '@core-ai/core-ai';

const result = await embed({
    model: mistral.embeddingModel('mistral-embed'),
    input: 'Search query for semantic similarity',
});

console.log(result.embeddings[0]);

Batch embeddings

const documents = ['First document', 'Second document', 'Third document'];

const result = await embed({
    model: mistral.embeddingModel('mistral-embed'),
    input: documents,
});

for (const [i, embedding] of result.embeddings.entries()) {
    console.log(`Document ${i}: ${embedding.length} dimensions`);
}

Code generation

const result = await generate({
    model: mistral.chatModel('codestral'),
    messages: [
        {
            role: 'user',
            content:
                'Write a Python function to calculate the Fibonacci sequence using dynamic programming.',
        },
    ],
});

console.log(result.content);

Provider-specific options

Options are namespaced under mistral in providerOptions:

Generate options

const result = await generate({
    model: mistral.chatModel('mistral-large-2512'),
    messages: [{ role: 'user', content: 'Hello' }],
    providerOptions: {
        mistral: {
            stopSequences: ['\n\n'],
            frequencyPenalty: 0.5,
            presencePenalty: 0.3,
            randomSeed: 42,
            parallelToolCalls: true,
            safePrompt: true,
        },
    },
});

Available fields: stopSequences, frequencyPenalty, presencePenalty, randomSeed, parallelToolCalls, promptMode, safePrompt.

Use parallelToolCalls to enable or disable parallel tool execution when using multiple tools.

Embed options

const result = await embed({
    model: mistral.embeddingModel('mistral-embed'),
    input: 'text',
    providerOptions: {
        mistral: {
            encodingFormat: 'float',
        },
    },
});

Error handling

import { ProviderError } from '@core-ai/core-ai';

try {
    const result = await generate({
        model: mistral.chatModel('mistral-large-2512'),
        messages: [{ role: 'user', content: 'Hello!' }],
    });
} catch (error) {
    if (error instanceof ProviderError) {
        console.error('Mistral API error:', error.message);
        console.error('Status:', error.statusCode);
    }
}

OpenAI Provider

GPT models with reasoning effort control

Anthropic Provider

Claude models with extended thinking

Embeddings Guide

Learn how to use embeddings

Core

Providers

Mistral Provider

Overview

Installation

createMistral()

Options

Returns

Supported models

Chat models

Embedding models

Provider support

Model capabilities

Strict tool schemas

Examples

Basic chat

Streaming

Reasoning output

Tool calling

Embeddings

Batch embeddings

Code generation

Provider-specific options

Generate options

Embed options

Error handling

OpenAI Provider

Anthropic Provider

Embeddings Guide

​Overview

​Installation

​createMistral()

​Options

​Returns

​Supported models

​Chat models

​Embedding models

​Provider support

​Model capabilities

​Strict tool schemas

​Examples

​Basic chat

​Streaming

​Reasoning output

​Tool calling

​Embeddings

​Batch embeddings

​Code generation

​Provider-specific options

​Generate options

​Embed options

​Error handling

​Related

OpenAI Provider

Anthropic Provider

Embeddings Guide

Overview

Installation

createMistral()

Options

Returns

Supported models

Chat models

Embedding models

Provider support

Model capabilities

Strict tool schemas

Examples

Basic chat

Streaming

Reasoning output

Tool calling

Embeddings

Batch embeddings

Code generation

Provider-specific options

Generate options

Embed options

Error handling

Related