跳轉到主要內容

Devstral Small 2

MistralMistral開源權重Apache 2.0 · 商用許可

描述

Devstral Small 1.1 (also called devstral-small-2507) is based on the Mistral-Small-3.1 foundation model and contains approximately 24 billion parameters. It supports a 128k token context window, which allows it to handle multi-file code inputs and long prompts typical in software engineering workflows. The model is fine-tuned specifically for structured outputs, including XML and function-calling formats. This makes it compatible with agent frameworks such as OpenHands and suitable for tasks like program navigation, multi-step edits, and code search. It is licensed under Apache 2.0 and available for both research and commercial use.

發布日期
2025-12-09
參數規模
24.0B
上下文長度
256K
支援模態
image, text

能力雷達圖

31
general
31
coding
37
reasoning
35
science
35
agents
80
multimodal

排行榜排名

領域#排名分數來源
程式碼能力榜324
34.0
AA
通用能力榜357
36.0
AA
科學能力390
34.0
AA

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

Code

SWE-Bench Verified53.6%自報

AA 評測指數

(Artificial Analysis)
Math Index(Artificial Analysis)
34.3
Coding Index(Artificial Analysis)
29.3
Intelligence Index(Artificial Analysis)
17.7
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))
0.7
Gpqa(NYU + Cohere + Anthropic (2023))
0.5
Livecodebench(UC Berkeley + MIT + Cornell (2024))
0.3
Aime 25(MAA (Mathematical Association of America))
0.3
Ifbench(Google Research (2023))
0.3
Terminalbench V2 1
0.3
Scicode(UIUC + Argonne National Lab (2024))
0.3
Lcr(Artificial Analysis)
0.3
Tau2(Sierra + U Toronto + Vector Institute (2025))
0.2
Terminalbench Hard(Stanford × Laude Institute (2026))
0.2
Tau Banking
0.1
Hle(Center for AI Safety + Scale AI (2025))
0.0

LLM Stats 分類評分

(LLM Stats (zeroeval))
Reasoning
50
Frontend Development
50
General
50
Code
50

定價

輸入價格免費
輸出價格免費
混合價格(3:1)免費

速度

Tokens/秒89.3
首Token延遲0.71s
首回答延遲0.71s

供應商價格排行

暫無提供商資料

外部連結