It's weird because two days after Jev was released there were a dozen decision models, a week later there are several dozen, mostly open source, OpenAI's own Decisions API [1] beats it, and you can easily finetune your own [2]. But as others have pointed out, this doesn't matter.

EDIT: As I wrote this Microsoft just released their own Decision-1 model [3].

[1] https://developers.openai.com/api/docs/guides/decisions

[2] https://unsloth.ai/docs/basics/train-your-own-decision-model...

[3] https://commandline.microsoft.com/microsoft-decision-1-model...

Everyone appears surprised by this news. It’s clear that they don’t have a product with some incredible moat. But they clearly have good engineering and product people that came up with a product people wanted. On top of that they have very strong marketing muscle that took the AI world by storm. And as far as I’ve seen, they still lead in some part of the latency-quality (-cost) curve?

They may well be a good team to throw money behind if you are hoping to bet on a new AI lab.

I really don't understand how this can be. I have sat in fund raising meetings with VCs in toronto and my experience is that there is shit ton of due diligence at the tech level. a product which has no moat, was already available, was duplicated within a couple of days is valued at 7B - i thought we were past the peak of the hype cycle.

Is Jev being astroturfed on HN? It certainly feels like it. It's a middling product with virtually no moat (but great marketing).

Jev does seem to have become the Kleenex of decision models. Is brand recognition worth $7.5B? There are lots of other decision models out there that perform at or near jev-level (laya, gliner 2.5 decide, even embedding gemma 2) that you can also run locally, and honestly I think this kind of model makes the most sense running locally as well. Maybe if TypeSafe can ship fast they can stay the default. Guess we'll find out.

Has anyone actually eval'd the other open source options against Jev on real world tasks rather than looking at benchmarks?

I see a lot of people parroting the quick open source alternatives as being better on the benchmarks, but it's such a new category that I'm not convinced we have solid benchmarks.

I'm hoping a company releases an internal eval benchmark for these options. I'm sure some of the open source ones are solid in some cases, but would love to see more reliable data.

Whether they can compete on decision models or not, TypeSafe showed that a lot of the market had missed something important. With this much money, they have a lot more chances to discover other important things that are missing.

No one finds it strange that every defector from Opensi gets half a billion dollars in funding and already has a third of Fortune 500 companies paying for their product in a few weeks with a product that's already been replicated 10 times very easily?

OK, I'm jealous... Lol

Every time I see these headlines I wonder why the Scala company is back in the news

Remember when reaching $1B valuation made you an exotic "unicorn"?

What edge do they have over the market to justify such evaluation

I wonder how much of it has been earmarked for astroturfing on HN and X? :D

I think this is a hedge against major AI regulation.

Invariably near-AGI systems created by OpenAI/Anthropic will be very destabalizing. In the end the world will probably regulate AI capable of [any] <-> [any] input/output types. Models will need to be limited on their outputs by law so they cannot have unbounded, unpredictable outcomes. Jev is the ideal version of "benefits of AI without making humans obsolete" that might be the consensus once the track superhuman AI and its consequences are clear.

Their headline says "TypeSafe A raises series AI". Is that AI slope?

Can someone who actually knows these things share how might a company like this spend $870M over the years?

Not a native English speaker; am I missing some kind of wordplay here or is the "I" in the wrong place in the title? It says: "TypeSafe A raises Series AI", but shouldn't it be "TypeSafe AI raises Series A"?

Hopefully some of the Typesafe hype spills over this way to Jeffy[0]

Open source[1]. It comes with 68 pre trained classifiers which run and train on CPU alone. They run locally, are faster than Jev/Laya/Decisions, and they can perform many different tasks; from labeling email, all the way to playing Doom[2]

[0] https://jeffyclassify.com/

[1] https://github.com/nicobrenner/jeffy

[2] https://playground.jeffyclassify.com/#doom

This might be the pets.com moment of this era.

There is significant congregation of investment towards companies like this whereas many other with meaningful innovation are struggling to survive. The model is not a revolution like I must say how llm did at the face value. Yes, transformers were there for long but the use case etc. If you do cost comparison using regular model and setting it with prompt and structured output you can get result very close. This is what most open Jev and other are doing with tuning llm.

It's funny how Andreessen Horowitz will go on a media blitz to ensure any company they invest in gets priority while Laya gets ignored. Hilarious.

Their job opening as Safety Javeroni was a good laugh too [0]

[0] https://jobs.ashbyhq.com/typesafe-ai/9a94651c-5d63-4e82-8854...

Good timing with our pac-man benchmark for decision models, now also including Microsoft Decision-1: https://smackernews.com/item/50007993 HN

I built JEV-search and JEV-chat and have been really happy with the results…

Curious to see how all of these can work together, and giving LLMs the ability to deploy their own workflows and built type 1 systems as they go has also been fun and useful.

Multi-hundred million raised at a multi-billion valuation for a series A is literally insane.

How are the investors hoping to get their money back? With a $20B acqui-hire from X.ai later this year or what's the plan?

They already got Sherlocked by OpenAI:

https://developers.openai.com/api/docs/guides/decisions

If you ask the right question, you don't need a decision model.

If you don't ask the right question, you might as well consult a psychic.

If all you need is scale, I guess decision models are cheaper than psychics.

Others have capably pointed out why their technology cannot be much of this valuation. I submit that “the ability to grab attention in the AI room” is what is being valued this highly.

What’s interesting about the Jev moment isn’t just Jev, it’s the unleashing of distillation / fine tuning outside the frontier-adjacent labs. It’s the sudden explosion of a million Jevs.

If being an “AI Researcher” is a ticket to multimillion dollar salary, AI training talent cannot be contained to a handful of companies. It’ll become more common and diffuse. The old advice of not fine tuning, because it’s hard, goes out the window as that knowledge diffuses through the industry.

A similar thing is happening in search. For a long time labs have trained tailored embedding models. And now companies like SID training their own agentic models that are smaller and faster at search than GPT-5.

This is pretty petty but they got the title wrong (should've been 'Typesafe AI Raises Series A' not 'Typesafe A Raises Series AI')

I only use Jev at the tool-calling layer to keep tool costs low, but I don't really see what the big advantage is.

Oracle needs to hurry up and get downgraded to junk status at which point it will raise shit and at which point it will finally send off the avalanche of normalization across this industry.

There needs to be cleansing with fire. Weeds need to die. Trees need their branches cut. The sooner the better.

A shitty ass random "ai" startup built entirely on hype and astroturfing should not be raising anywhere near this amount of money.

What a neat API.

I went ahead and applied for a job, and unlike Oxide, TypeSafe has an application that only takes a few minutes to fill out.

I did take a second to play with the TypeSafe API first. I'm also excited to get Jev running locally.

Not like I'm realistically good enough for a top AI company, but I can dream. I got multi code signal tests for Anthropic and got 70% at best.

Would be nice to get equity, ipo and cash out.

I don't want to work in my 50s

interestingly , it destroy the landscape of Chinese models.

Unless china takes leadership in frontier space the picture is next :

1. cheap workhorses for classification, routing, other scenarios : Jev 2. coding agents with less erros : Anthropic/Openai, etc. 3. Science /Legal/Medical : A mixture of Jev+Anthropic scenarios

If Typesafe really wants to do some good in the world, it'll raise money for those in need.

That press release made me cringe a bit. Maybe corporate speak wasn't so bad after all.

The amount of investment being thrown at this when we have serious existential problems as a species is how I know the system is definitively broken. Capitalism got us this far, but it's time for something new.

Do they have patents over jev related tech or something valuable to justify this?

This looks like the top of the dot-com bubble...

I guess they are investing in the team rather than on their demo product

Tbf this is THE way to get cost-effective AI into video games (think enemy AIs and NPC AIs that are much better than current ones and runnable on consumer GPUs), i'm really looking forward for decision models maturing.

got the recruiter call only to essentially be summarily rejected because my pedigree is wack. looks like i would've gotten hosed on valuation anyways.

~7B for a thing that we already have open-source?

A few year ago you could IPO at this valuation.

These investors are not that smart

I’ll just use open source, thanks

So... are they worth more, or less than 0xide

Only just now I realized that "TypeSafe AI" are the people behind Jev, and "System One" isn't the company name, as I assumed, but a larger project label.

This makes no sense, unless they are going to crack something, like the actual atrocious accuracy of around 70%.

It's always been the last 30% where it just fails.

Guess it is 870M to 0

Another data point confirming that AI is a bubble.

From Andreessen the Trump donor. Sad, will switch to something else now.

Good job typesafe.

You won the competition with VCs

Lol

this market lol

Disclosure: I work at H2O.ai.

We released an Apache-2.0, open-weight 4B decision model that scores above Jev 1.13 on JevBench's composite score (72.5 vs 71.5) and is currently the top open model there: https://benchmarkheaven.com/jev-models . Newer models coming even larger than beat Jev in intelligence as well.

- Same contract as Jev: state + typed questions in, calibrated probabilities out, one forward pass, no generated tokens. - Your data never leaves your environment, and there's no per-call fee.

Weights, card and run instructions: https://huggingface.co/h2oai/h2o-lightning-4b

Is there a second wave of AI bubble happening? How can an AI classifier company be worth of $7.5B