I also went through quite a process to select a language to work with LLMs[1] before settling on OCaml.
I am not unhappy with the choice and find it works quite well, with relevant skills loaded, but I am always interested in others’ experience and understanding what they’ve discovered works well.
I received the same email, although couldn’t quite figure out which retiring model I was still using, as I thought I’d already transitioned to Mistral-Medium-3.5 for everything. Anyway, after receiving the email, my hope was that it meant they were also planning on releasing some new, improved models in the next months.
Make sure the specifications can’t fail by verifying them for correctness.
Something like TLA+[1] and Quint[2] specifications can be verified for correctness using Apalache[3]. Then test the Rust code against the specifications using quint_connect.[4]
I was a Mistral Le Chat Pro subscriber (the €20/month plan). Yesterday I hit my monthly limit. Switching to PAYG I burned through another €40 in one evening, working on the same project, with the same tasks.
I upgraded my plan last night to Mistral Le Chat Teams. This now costs me €60 per month for two users. Limits have been reset, but I have no idea now if my per seat limit is higher than the Pro plan, or if the limit is shared between the seats, it’s really not clear. I guess I will find out next month. The limits reset on the first of the month and I really hope I don’t hit them in the next seven days.
I use Mistral Vibe CLI and I’ve written and implemented a couple of new skills[1]. Caveman, based on an idea I found online somewhere, this skill removes all extraneous response text, including articles. Makes for some fun reading, but supposedly reduces output tokens significantly. Hash-anchors, this one is based on a concept from Dirac[2], reduces search failures and also includes multi-file dispatch. It will be hard to measure, but Vibe tells me these two should result in roughly a 40% reduction in token burn.
It's generally inconsistent. The first sentence is written, "A co-op is an economic system built on the simple idea that coordinating the economic activity..."
Co-op is correct here, but not in the title (Coop). Probably personal taste, but I'd also like to see hyphentation for "co-ordinating", "co-operate" and "co-ordinator" as well.
Then I noticed the em-dashes, so perhaps I'm reading the machine's work anyway.
I don't use the numpad characters, but I have tried Vibe, Goose (GUI and CLI), Dirac and the built in agent in Zed for vibe coding. I keep coming back to Mistral's Vibe. I actually find the ergonomics of it nicer than the others I've used so far. I really wanted to like Goose, and their GUI offering is OK for chat, but I thought their CLI was poor. Dirac was OK and I should try it again to be fair. Zed was just overkill and complex for what I needed. Vibe CLI seems to hit the sweet spot, although it's not perfect. The challenges I encounter are mostly down to API errors though and sometime bash tooling. I could configure it better for that, if I took the time (which I should).
Of course ChatGPT is available to EU citizens, should they choose to use it. That’s very different to the Maltese government actively promoting use of ChatGPT.
Of course Maltese citizens can still choose not to use ChatGPT (until it becomes mandatory), but if the State supported education is bound to one particular tool, storing user data outside the EU’s jurisdiction, I think that’s something to discuss.
Malta is part of the EU. I am personally very surprised about this partnership, just in the context of data security, privacy and the GDPR. How is the privacy of these EU citizens protected when all their prompts and data is sent to OpenAI? How do these EU citizens submit a request for all their personal data to be deleted from OpenAI records, a right they have under the GDPR with a compliant data processor?
Also in this space is Moonbit[1] and Vera[2]. A performance comparison between them, across multiple factors, would be interesting because I have no idea of the pros and cons of each.
The initial criteria was strongly typed and functional first. Using an LLM for answers, of course, that returned me a list that looked like:
- Haskell
- OCaml
- F#
- Scala
- Gleam
- Purescript
- Grain
- Idris
Then I asked if there were any Schemes or Lisps that met the initial requirements, which added a bunch more options (Typed Racket, Typol, Elm, ReScript etc).
Then I asked about Julia specifically, as it's a language I'm already reasonably familiar with and knew that it's possible to write it with static annotations.
Next I started filtering the list based on additional criteria; didn't want to target a JS compilation target, performance, size of package ecosystem, tooling, community, learning curve (I do want to review and understand the output).
There were a bunch of follow-up questions over a few hours of prompting, reading and a couple of beers. All this resulted in the shortlist of OCaml, Typed Racket and Julia.
Julia pretty much remains in there, even though it doesn't really meet the strongly typed initial criteria, based on my familiarity, the ecosystem especially for AI/ML tasks and performance factors.
I know zero about OCaml and find the thought of learning it a bit daunting. Typed Racket seems more approachable anyway.
I also went through quite a process to select a language to work with LLMs[1] before settling on OCaml.
I am not unhappy with the choice and find it works quite well, with relevant skills loaded, but I am always interested in others’ experience and understanding what they’ve discovered works well.
[1] https://news.ycombinator.com/item?id=48105850