Devstral Medium

Mistral AI 與 All Hands AI 合作開發的高階程式碼智能體模型，專門用於軟體工程任務。在 SWE-Bench Verified 基準測試中達到 61.6% 的成績，超越 Gemini 2.5 Pro 和 GPT 4.1，但價格僅為其四分之一。支援企業級部署和自訂微調，適合需要高性能程式碼生成和多檔案編輯的複雜開發場景。

Intelligence(Relatively Weak)

Speed(Relatively Slow)

Input Supported Modalities

Is Reasoning Model

256,000

Context Window

128,000

Maximum Output Tokens

Knowledge Cutoff

Go Compare

Pricing

￥2.88 /M tokens

Input

￥14.4 /M tokens

Output

￥5.76 /M tokens

Blended Price

Quick Simple Comparison

Input

Output

Devstral Small (May '25)

￥0.1

Codestral (Jan '25)

￥0.2

Devstral Small

￥0.1

Basic Parameters

Devstral MediumTechnical Parameters

Parameter Count

Not Announced

Context Length

256.00k tokens

Training Data Cutoff

Open Source Category

Proprietary

Multimodal Support

Text Only

Throughput

137

Release Date

2025-07-10

Response Speed

97.926 tokens/s

Benchmark Scores

Below is the performance of Devstral Medium in various standard benchmark tests. These tests evaluate the model's capabilities in different tasks and domains.

Intelligence Index

Large Language Model Intelligence Level

38.18

Large Language Model Intelligence Level

Coding Index

Indicator of AI model performance on coding tasks

31.52

Indicator of AI model performance on coding tasks

Math Index

Capability indicator in solving mathematical problems, mathematical reasoning, or performing math-related tasks

38.7

Capability indicator in solving mathematical problems, mathematical reasoning, or performing math-related tasks

MMLU Pro

Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video

70.8

Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video

GPQA

Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions

49.2

Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions

HLE

The model's comprehensive average score on the Hugging Face Open LLM Leaderboard

3.8

The model's comprehensive average score on the Hugging Face Open LLM Leaderboard

LiveCodeBench

Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems

33.7

Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems

SciCode

The model's capability in code generation for scientific computing or specific scientific domains

29.4

The model's capability in code generation for scientific computing or specific scientific domains

HumanEval

Score achieved by the AI model on the specific HumanEval benchmark test set

93.5

Score achieved by the AI model on the specific HumanEval benchmark test set

Math 500 Score

Score on the first 500 larger, more well-known mathematical benchmark tests

70.7

Score on the first 500 larger, more well-known mathematical benchmark tests

AIME Score

An indicator measuring an AI model's ability to solve high-difficulty mathematical competition problems (specifically AIME level)

6.7

An indicator measuring an AI model's ability to solve high-difficulty mathematical competition problems (specifically AIME level)

Output tokens/million

Output tokens/million

Output tokens/million

Output tokens/million

Output tokens/million

131k

Context Length

Gemini 2.0 Flash Lite (Preview)

Output tokens/million

Output tokens/million

Output tokens/million

8192

Context Length

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご