frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Mistral Large 4

https://docs.mistral.ai/models/mistral-large-4-0
140•funnym0nk3y•50m ago

Comments

moecables•30m ago
compared to Deepseek 1017, I'm not sure why I'd use this?

https://openrouter.ai/compare/mistralai/mistral-large-4-0/de...

serioussecurity•28m ago
If you're required to use a European model
walrus01•25m ago
Interestingly Mistral has been selling GLM5.3 inference hosted in Europe on their own hardware, for those who are OK using a "not developed in Europe" model but run on hardware located in and controlled by a European corporation.

https://docs.mistral.ai/models/zai-glm-5-3

frotaur•23m ago
Indeed, I see no difference between a chinese model hosted in Europe vs a Europe-grown model.

You could argue that potentially you can train a backdoor into a foreign-grown model (that has been demonstrated to be possible in toy examples), but I would highly doubt that this would be the case today, given the potential reputational damage if discovered.

debugnik•15m ago
Didn't we get a submission today about Qwen3-Coder producing more bugs when hinted on the user having different ideology than the Chinese government?

I see it's flagged now, probably fake, but the fact that it's possible does make me feel uneasy about both Chinese and US models.

walrus01•12m ago
I recall seeing something about that. I don't know of anyone still using such an ancient model when much newer Qwen 3.5-120, 3.6-35B-A3B, 3.8-27B and 3.8-flash-next all exist and score much higher on all conceivable metrics. I would like to see if anyone can replicate the same thing on a 'current' Qwen released in the past 6 months.
debugnik•9m ago
Agreed, but even as an unintentional bias I'd find this problematic for us.
verdverm•9m ago
It was so insignificant, it could be equally described by the non-determinism inherent in the models. And that it reads more like a marketing piece.

That's my read anyhow. It's been flagged. The whole thing is kind of rich, considering the Israeli roots of the company.

https://news.ycombinator.com/item?id=50179706

IshKebab•9m ago
> I see no difference between a chinese model hosted in Europe vs a Europe-grown model.

Really? Fascinating. Do you know about Tiananmen Square, the great firewall of China, etc?

ex1fm3ta•23m ago
if you care about data-retention policy or you have to use european models ...
Systemerror7A69•23m ago
Well, Models are hard to just distill down to a single benchmark number, or even a few across multiple different benchmark - they all behave and work slightly differently and you can find some who work or feel better for you and your use-case specifically.

But you're right, Mistral Large 4 seems a much more expensive but overall weaker model across the board.

The reason to use them would be, well, because they're...Mistral. They're in EU, france is large in nuclear energy so the energy cost of training has much less of an environmental impact, the data protection laws are very good in EU. This can really be a deciding factor for many EU companies, and I think it's good that they now have something which is at least semi-usable. (apparently)

It's also good that other options exist, even if they are worse. Deepseek is kind of like a large specter looming over every model release, because it's just SO cheap and good. That doesn't mean we should abandon every other model.

Still, the model is not really competitive on price and abilities alone. I'm mostly happy to see Mistral still in the fight.

petercooper•19m ago
I say this with a slight tongue in cheek, but Mistral does have a better option: they recently added GLM 5.3 to their platform (and it works rather well). Not trained in the EU, though, of course ;-)
andix•19m ago
Patriotism.
lwansbrough•7m ago
Well when your greatest ally acts like an adversary, who can blame them for the increased sovereign focus?
yorwba•28m ago
Previous discussion: https://news.ycombinator.com/item?id=49977979 (2044 points 15 days ago, 1254 comments)
Flux159•24m ago
I think that this link is more helpful: https://mistral.ai/news/mistral-large-4/

Seems like it's multimodal and visual intelligence capabilities are better than Deepseek 4.1 Flash, even though coding is not equal to DS 4.1 1017. The cost is about 10x of Deepseek though comparing on OpenRouter.

Weights will be here https://huggingface.co/mistralai/Mistral-Large-4-1T-A52B at the end of the month.

Systemerror7A69•19m ago
It's gonna be interesting to see the final version, since this is a preview model still. I'm not completely sure as to how this works, but from what I gathered it will be further trained during this period, so it might improve a bit.

Since cost is so much higher than deepseek though it really seems like the biggest reason to use it would be "in the mistral subscription" for personal use or "because you need to abide by european data privacy laws" for companies.

Mostly I'm happy they're still in competition, as I mentioned already elsewhere.

andix•20m ago
I see Le Chaton Fat and press upvote, sure. But why? Wasn't it released already 2 weeks ago?
wojtek1942•11m ago
Some details and benchmarks were announced but the weights didn’t get released
Betelbuddy•9m ago
Le Chaton Fat is a lovely name. I predict the next Michelin restaurant with will have an LLM as the Chef...
nik736•4m ago
The weird thing about this model is that it spits out Chinese characters just like the chinese models when chatting with it in German. Happened to me quite often during the past few days.
dropbox_miner•3m ago
waiting for pelican bench results (the only benchmark that matters)
kolinko•5m ago
Remember Stuxnet? An american worm that was completely benign, travelled on usb drives in hiding, and only activated when it guessed that it's on Iranian plants, and iirc changed very subtly some very specific parameters of operations.

Here, with a chinese model, it would be even simpler - you finetune a model to activate certain operations when it discovers certain environments, but to behave completely normal otherwise. Kind of like a sleeper agent. And with current tech we wouldn't even be able to spot this.

GT3 racing game in the browser with WebGPU

https://webgpu.com/showcase/project-endurance-webgpu-racing/
1•FarhadG•1m ago•0 comments

Show HN: I made a free Mac app to annotate screen for AI agents

https://www.trylasso.online
1•nikasakana•3m ago•0 comments

Ask HN: Can we immortalize your startup in our Ultracapitalism game?

2•ChromaGolem•3m ago•0 comments

Meetings [pdf]

https://www.nber.org/system/files/working_papers/w35706/w35706.pdf
1•neehao•4m ago•1 comments

CDC Ignored Request from States to Validate a New Measles Test

https://www.nytimes.com/2026/10/21/health/measles-test-cdc.html
4•gumby•6m ago•0 comments

Show HN: Twigg Links – go/ links defined in a file in your repo

https://twigg.vc/docs/v/2/blog/twigg-links
1•mark-glue•7m ago•0 comments

Who's Winning AI, a live scoreboard for the AI race

https://whoswinningai.com/
1•mielve•9m ago•1 comments

Ford to City: Drop Dead (1975)

https://en.wikipedia.org/wiki/Ford_to_City:_Drop_Dead
2•bryanrasmussen•10m ago•1 comments

4.7% of Open Source Projects Have One or Two Maintainers

https://hagen.rip/articles/4.7%25%20of%20Critical%20Open%20Source%20Projects%20Have%20One%20or%20...
2•APoorStudent•10m ago•0 comments

AI Early adopters share their views on AI status

https://automato.substack.com/p/where-is-ai-heading-in-october-2026
1•josmor•10m ago•0 comments

NOAA fabricates data for more than 30% of the 1,218 stations that dont exist

https://wattsupwiththat.com/2024/04/12/hidden-behind-climate-policies-data-from-nonexistent-tempe...
1•bilsbie•11m ago•0 comments

Multicore TypeScript, Inspired by Rust's Rayon

https://geastack.com/blog-parallel-typescript
1•arbayi•12m ago•0 comments

Do minor curl vulnerabilities in the open

https://daniel.haxx.se/blog/2026/10/20/do-minor-curl-vulnerabilities-in-the-open/
1•birdculture•12m ago•0 comments

AI Portability in 2026: Breaking Up Is Hard to Do

https://peckjon.substack.com/p/ai-portability-in-2026
1•peckjon•13m ago•0 comments

The AI-Hacking Apocalypse Began Last Night

https://www.theatlantic.com/technology/2026/10/ai-hacking-apocalypse-began-last-night/689106/
4•zzzeek•15m ago•1 comments

Tinder Reverse Engineered

https://twitter.com/TheDonAyo/status/2112591782857109809
2•mrkn1•19m ago•0 comments

Ollama flint requires 0.40.8 but most recent is 0.40.7

1•roscas•20m ago•0 comments

Show HN: Yantra – Interactive step-by-step diagrams as one offline HTML file

https://github.com/niyankhadka/yantra
1•niyankhadka•21m ago•0 comments

Developers Are Dumbing Down Their Smart TVs

https://www.extremetech.com/electronics/developers-are-dumbing-down-their-smart-tvs
2•speckx•23m ago•2 comments

NPWS ask for vigilance after red kite poisoning in Carlow

https://www.rte.ie/news/regional/2026/1021/1596338-red-kite-carlow/
1•austinallegro•24m ago•0 comments

After automation: Solo founding will be the default

https://every.to/thesis-statements/julian-weisser
4•rmason•26m ago•2 comments

Ask HN: Caution on using AI security platforms for repositories.

1•BehanPrW•27m ago•0 comments

Claude Duet

https://claude.com/resources/articles/claude-duet
6•zaevlad•30m ago•2 comments

Physicists trap a single antimatter particle for a month

https://www.scientificamerican.com/article/physicists-trap-a-single-antimatter-particle-for-a-month/
1•mooreds•30m ago•1 comments

Research Preview of Claude Duet

https://twitter.com/claudeai/status/2112984664256241837
1•davidgomes•30m ago•1 comments

How to Fix PHP2Shell (CVE-2026-109618 and CVE-2026-109619)

https://containersecurity.dev/blog/how-to-fix-php2shell
1•aerabi•33m ago•0 comments

Vulnerabilities in PHP Affecting Cisco Products: October 2026

https://sec.cloudapps.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-php-octobe...
1•speckx•33m ago•0 comments

4 Wheel Steering Is a Nightmare

https://simulation.michelin.com/canopy/technical-articles/4-wheel-steering-is-a-nightmare
1•shpx•34m ago•0 comments

iPhone Duo Review with CT scan [video]

https://www.youtube.com/watch?v=o1qc29t04GI
1•tekacs•35m ago•0 comments

Plex laid off the whole player team

https://forums.plex.tv/t/2026-20-0-beta-ios-and-tvos-4k-dovi-hdr10-won-t-play-infinite-spinning-c...
49•zzyzxd•38m ago•27 comments