{"id":35799,"date":"2025-05-22T12:34:30","date_gmt":"2025-05-22T04:34:30","guid":{"rendered":"https:\/\/www.1ai.net\/?p=35799"},"modified":"2025-05-22T12:34:30","modified_gmt":"2025-05-22T04:34:30","slug":"mistral-%e5%8f%91%e5%b8%83%e7%bc%96%e7%a8%8b%e4%b8%93%e7%94%a8%e5%bc%80%e6%ba%90-ai%e6%a8%a1%e5%9e%8b-devstral%ef%bc%9a240-%e4%ba%bf%e5%8f%82%e6%95%b0%e3%80%81%e5%8d%95%e5%bc%a0-rtx-4090-%e6%98%be","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/35799.html","title":{"rendered":"Mistral Releases Devstral, an Open Source AI Model for Programming: 24 Billion Parameters, Runs on a Single RTX 4090 Graphics Card"},"content":{"rendered":"<p>Technology media outlet WinBuzzer published a blog post yesterday, May 21, reporting that Paris, France-based <a href=\"https:\/\/www.1ai.net\/en\/tag\/mistral\" title=\"[See article with [Mistral] label]\" target=\"_blank\" >Mistral<\/a> AI companies for advanced software engineering tasks.<strong>roll out<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e5%bc%80%e6%ba%90\" title=\"[View articles tagged with [open source]]\" target=\"_blank\" >Open Source<\/a> AI model <a href=\"https:\/\/www.1ai.net\/en\/tag\/devstral\" title=\"_Other Organiser\" target=\"_blank\" >Devstral<\/a>.<\/strong><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-35800\" title=\"6925d1d0j00swnbci0083d000to00gpp\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/05\/6925d1d0j00swnbci0083d000to00gpp.jpg\" alt=\"6925d1d0j00swnbci0083d000to00gpp\" width=\"1068\" height=\"601\" \/><\/p>\n<p>The model, developed by Mistral AI in collaboration with All Hands AI, has a parameter size of 24 billion, is currently in \"research preview\" and is released under the Apache 2.0 open-source license, which allows developers and companies to use it for commercial purposes without restriction, according to a blog post cited by 1AI.<\/p>\n<p>In an official blog post, Mistral AI emphasizes that Devstral is driving the evolution of \"agentic\" coding, where AI takes a more proactive role in the development process, not only generating simple code snippets, but also tackling complex development challenges in the real world.<\/p>\n<p>Developers can access the model (codename devstral-small-2505) via the Mistral API, priced at $0.1 per million tokens in and $0.3 per million tokens out; the model also supports downloads from platforms such as Hugging Face, Ollama, and others for local deployment.<\/p>\n<p>Devstral has performed well in performance tests. According to Mistral AI, the model scored 46.8% in the SWE-Bench Verified benchmark, outperforming the previous state-of-the-art open-source model by more than 6 percentage points, and even surpassing some of the closed-source models such as OpenAI's GPT-4.1-mini.<\/p>\n<p>Devstral runs smoothly on a Mac with a single Nvidia RTX 4090 graphics card or 32GB of RAM, dramatically lowering the hardware barrier.<\/p>\n<p>Mistral AI says Devstral specializes in exploring codebases, editing multiple files, and supporting code agent frameworks like OpenHands and SWE-Agent to solve real GitHub problems.<\/p>","protected":false},"excerpt":{"rendered":"<p>Technology media outlet WinBuzzer published a blog post yesterday (May 21) reporting that Paris, France-based Mistral AI has launched Devstral, an open-source AI model for advanced software engineering tasks. Citing the blog post, 1AI said that the model was developed by Mistral AI in collaboration with All Hands AI, has a parameter scale of 24 billion, and is in the \"research preview\" phase. It is currently in the \"research preview\" stage and is released under the Apache 2.0 open source license, which allows developers and companies to use it for commercial purposes without restrictions. In its official blog post, Mistral AI emphasizes that Devstral promotes \"agentic\" coding, i.e., AI that is not a part of the development process, but a part of the development process.<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[167,6710,559,219],"collection":[],"class_list":["post-35799","post","type-post","status-publish","format-standard","hentry","category-news","tag-ai","tag-devstral","tag-mistral","tag-219"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/35799","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=35799"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/35799\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=35799"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=35799"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=35799"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=35799"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}