Even as someone happy of seeing Mistral still trying to compete, I would wait at least for the final release and weights before talking again about it, to have a definitive opinion on it.
I have never understood why people using MicroG can use apps like SNCF-Connect, while I cannot.
SNCF-connect is the largest online marchant in France. Since june, I cannot click the "pay" button anymore. It coincides with the new feature of Pay with Google.
So : is it because MicroG allows calls to Google services on e/OS devices, while my LineageOS refuses ? Or because MicroG implemented a try catch around this call, without any call to Google's servers ?
If it connects to Play Store, then isn't that the same as Aurora? Can people's accounts be banned? That would really suck to pay for something and lose it because Google banned you haha!
It really sucks but there is no other widely supported standard for in-app purchases. Some services refuse to use credit/debit cards payments for some reason.
Some Play Services calls are stubbed out as no-ops and others are reimplemented either with fully free replacements (e.g. navigation) or else by calling proprietary Google remote endpoints (e.g. firebase)
Try the per-app exploit protection compatibility mode. That's generally the first thing to try, especially for apps with dubious anti-tampering code. If it works then it can be narrowed down to the specific exploit protection and should usually be reported to the app developers. The default enabled exploit protections for user installed apps have no false positives and neither does memory tagging (MTE).
I've been using Mistral Large 4 and some of my other friends have as well.
Something is going on with the performance (as in speed of the service, not quality). Some days it's quite fast, some days it's so slow I interrupt it because I thought it was hanging.
I think they're shifting hosting capacity around, or otherwise messing with the model. Note that it's still a "preview" version, meaning they're still training it further for the final release and they are likely still figuring out the optimal hosting configuration as well.
It wouldn't surprise me to see further changes in both quality of the model and speed before the final release.
I'm looking forward to the weights release as well, has anything about the architecture and possibly quantization resilience been shared publicly so far?
I updated my vibe config to point to Mistral 4 and it charges my subscription capacity, not my API allowance. So it is included in subscriptions, you just need to do a little bit of legwork.
It was for me too, and it's only become slow very recently. I can't say with certainty, but I think it was around the time Mistral 4 launched that it seemed to slow down.
Odd, I am in France too. Progressive rollout maybe? I find it weird to announce a model with fanfare but not make it immediately available to customers, let alone paid ones.
> keeping it up-to-date with its dependencies etc.
not if the software is functional today, and doesn't require interacting with third party systems (like this tinylittleoffice suite).
By the time the underlying infrastructure (like browser compatibility, or self-hosting OS becoming obsolete) fails, you're likely no longer using it anyway.
The "documents" seem exportable in text format, and in any case, potentially salvageable with another vibe-coded tool if such need arise.
Therefore, the original author is no required to maintain it.
That's a rosy eyed view. users running outdated code not getting security fixes are sweet honeypots for modern automated llm-backed mass exploitation machine long before something obviously fails for you.
you'll be spending tokens on getting it maintained, and if you don't spend enough you get pwned by guys who do. who exactly is winning in this scenario?
OSS maintainedness should not at all be taken for granted. If you need something maintained by pros and security experts, that more accurately describes commercial software. FOSS as it stands, with the exception of some (but nowhere near all) major lynchpin projects, is done by volunteers in their spare time.
A lot of OSS gets supported by foundations, selling consulting or additional services, donations, maintainer's dayjobs, and people's pride in their work and desire to help
I am placing your time of digital fruition in the 00’s or later? Not bashing, but we bought floppies in the 80s with software with no updates. Cool kids got their software from Oh No More Warez + keygen in the early 90s. That went online in the late 90s.
Always online, internet of shit and perma update crap came at that same time, first for the OS and later for everything. Oh, plus two decades of using shady Russian software for virus scanning. Shit went up, update everything became everywhere and everyday. Plus gigabyte frameworks for functionality that was kilobytes before.
My point in cursing too much? (Sorry, mentally screaming FUD here.) That tiny code projects, even without an active maintainer, in the LLM-era are probably a lot more secure when run by the people they target (we run containers, and sandboxes, the cool kids run immutable distros). This is democratizing software, and it has the great feel of old.
Please everyone, vibe code tiny project with tiny codebases everyday. Dump them on GitHub in a free tier. Make so many that the people have a free choice again! And go watch Cliff Stoll again for that hacker energy!
Spending tokens to search for vulnerabilities in commonly used software makes sense because the payoff is large - hack once, exploit many.
However for unique, bespoke software (vibe-coded or not), the payoff is as small as a single endpoint. You’ve found an exploit in MikesOwnOfficeSuite and now have access to Mike’s PC. You’ve got to spend the same tokens again to find an exploit in SuesAwesomeOfficeApps. It still might be worth it in a spear-phishing scenario.
if everyone vibe coded their own offices it would be a heaven for big tech llm operators lining up their pockets of course, but the idea was that Mike is using something vibe-coded by somebody else. just target TinyLittleOffice and get everyone using it
This is the part managerial and marketing types don't understand. Their expectation is usually that "since its a computer system (sic) the tech team will support it from now on".
patching dependencies correctly was one of the first agentic coding tasks llm solved 100%. you will need to dedicate a session once a year to to such patches. Or community of vibe coders will do it for you .
That has always been the case with anything self-hosted. Otherwise you're putting a lot of faith in an upstream maintainer to actually put in the effort which they may or may not be doing.
reply