Devstral
Fewer than 5 people recommend it so far, so there's no written verdict. Here is what commenters said.
2recommend
4warn against
Dec 9, 2025first mention
May 10, 2026latest mention
Why people recommend it
“I've been using devstral, codestral and Le Chat exclusively for three months now.”
Caveats and dissent
“it hallucinated so much that it wasn't super helpful”
“My biggest issue with Devstral and even their biggest model is that they’re dangerous unless closely directed and reviewed”
Show 2 more
“I feel sorry for whatever they did to that model”
“it worked extremely poorly with the tool calls”
Often compared with
Threads where it came up
- item 47431671 3 mentions
- Task Paralysis and AI 1 mention
- France's Mistral Built a $14B AI Empire by Not Being American 1 mention
- Running Local LLMs Offline on a Ten-Hour Flight 1 mention
- Lambda Calculus Benchmark for AI 1 mention
- Ask HN: What is the best software IDE to program with a local LLM and GPU? 1 mention
Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 13
“it hallucinated so much that it wasn't super helpful”
“My biggest issue with Devstral and even their biggest model is that they’re dangerous unless closely directed and reviewed”
“It was ok-ish.”
“even local ones like devstral”
“I definitely remember using copilot with devstral”
“The Devstral logical deduction result (0.22→0.76) was a surprise to me.”
“Devstral was the first one I tried”
“I published the results for devstral”
“I feel sorry for whatever they did to that model”
“I've been using devstral, codestral and Le Chat exclusively for three months now.”
“I am just starting to experiment with Claude and Devstral”
With RAM prices spiking, there's no way consumers are going to have access to frontier quality models on local hardware any time soon, simply because they won't fit. That's not the same as discounting the open weight models though. I use DeepSeek 3.2 heavily, and was impressed by the Devstral launch recently. (I tried Kimi K2 and was less impressed). I don't use them for coding so much as for other purposes... but th
“it worked extremely poorly with the tool calls”