Devstral 2

“it hallucinated a few discrepancies but also found real stuff i missed in the docs”

“LLMs are generally good at following a specific style when given examples.”

“Devstral is getting better, it's the Vibe harness that's holding it back”

“I'm now using the way too heavy "devstral" for these tasks.”

“it hallucinated so much that it wasn't super helpful but it still shot out boiler plate code”

“Devstral Small 2 however just works, for the most part.”

“It was ok-ish. Lots of hand holding, figured out I can't use it for planning much”

“After testing Le Chat and Devstral-2 for a while, I felt their offering was good enough to stump up some cash for it.”

“They can all do some tasks reliably, while struggling at other tasks.”

“I'm a huge fan of local models, have been using them regularly ever since devstral1 released”

“it will cut corners and write incomprehensible code even with a good Opus plan to boot”

“I've been using devstral2 with great success for a few months now.”

“right now the perceived best model is Devstral-Small-2-24B-Instruct-2512-4bit for me”

“a dense model like Devstral2 24B slaps on the Dual GPU setup”

“The model pointed way better in gsm8k, but lost a bit on the other categories.”

“I used the devstral with the vibe.cli and it look sharp to me”

“Recent Devstral 2 (mistral) is pretty precise and concise in it's changes.”

“all running on my own old PC with everything fitting in a 4 euro USB stick”

“which is not really open source but more of a "MIT unless you have lot of money"”

“I haven't accepted one Claude, devstral or gpt suggestion verbatim”

“They are in the top of open models, and surpass some closed models.”

“it is insane to me that such a small model can follow those explicit directions and consistently perform that workflow”

“The current SotA for open models (dsv3, glm, minimax, devstral, etc) are at or above the mini versions of top labs”

“The best you can get today with consumer hardware is something like devstral2-small(24B)”

“looks promising as something I want to try fine tuning on DSLs, etc.”

“simple tasks CAN be handled by Devstral 24B or Qwen3 30B A3B, but often they fail at tool use”

“I use DeepSeek 3.2 heavily, and was impressed by the Devstral launch recently.”

“The price to performance ratio is practically better in every way.”

“I have high hopes for Mistral's Devstral 2 but I've not run that locally yet.”

“A 123B dense code model under a permissive license plus an open-source CLI agent”

“it's nonetheless good enough and doesn't require me to send my data off my laptop”

“it did ok finding issues but completely failed doing rather basic edits, like it lost closing brackets or used wrong syntax and couldn't recover”

“I can see myself shifting to this for reguar day-to-day coding if they can offer this for copetitive pricing.”

“it's impressive. It walked through a data analytics and transformation problem”

“Finally, we can use a european model to replace claude code.”

“Somehow it writes bad React code and misses to check linting prompts half the time.”

“for coding that's a great approach given how much lead in coding performance Qwen and Deepseek have”

“only restrict companies with massive revenues from using them for free”

“Devstral 2 ships under a modified MIT license... doesn't look particularly permissively licensed (or open-source)”

“one of the better ones to run locally for agentic coding”