ChatGPT broke Diogo AlmeidaâÂÂs heart.
Almeida was an OpenAI researcher who helped build the chatbot and then invent reinforcement learning from human feedback (RLHF), the model-training technique perhaps most responsible for our current age of AI. But despite its capabilities, he was disappointed.
âÂÂWe have lightning in a bottle, and yet it is not useful,â Almeida told TechCrunch. âÂÂIâÂÂve been battling that problem since then. It took me a while to come to the conclusion: the problem is we are optimizing for human language ⦠We have been super good at human language for four years, but itâÂÂs not useful for automation because computers speak a different language.âÂÂ
Two years ago, Almeida left OpenAI to start TypeSafe AI, a startup trying to fix that problem. This week, the company released a new transformer-based model, Jev, that is not a large language model (LLM). It doesnâÂÂt output text, but instead produces probabilities, or what the company calls âÂÂcalibrated decisions.âÂÂ
Eschewing language does a few things: It makes the model incredibly cheap and fast, and because users define the outputs in advance, it cannot hallucinate. Its output tokens are free, and input tokens are metered by the billion, not the million.
Developers are taking a great interest in the product; the company briefly lost the ability to serve users from its API because demand was so high. Jev appears most useful for software automation. Thus far, software developers see it as a cheaper and more robust way to incorporate intelligence into their code.
For example, Pranit Sharma, a software engineer at Vercel, a company making agentic infrastructure, said his company had used OpenAIâÂÂs ChatGPT Luna 5.6 to run a classifier to review commands for safety. When Vercel replaced OpenAIâÂÂs Luna with Jev, it got results 5 to 18 times more quickly and with greater accuracy.
Another developer, Bryo AI CTO Nikhil Mudholkar, tested Jev against Gemini for classifying business emails. In his test, Gemini was slightly more accurate, but 10 to 20 times more expensive. More interesting to Mudholkar were JevâÂÂs confidence scores â âÂÂit is the only one that hands back a real probability which makes it ideal for automating workflows!!âÂÂ
Besides replacing LLMs in certain use cases, the new model can also augment them, acting as a smart check on misbehavior. Using agents to monitor agents can quickly become expensive, but using Jev to do so, Almeida argues, makes sense. He sees users deploying Jev to track LLM agent traces and prevent jailbreaks.
âÂÂAt the end of the day, it delegates the hallucination problem a little bit to the user,â explained Armin Ronacher, the CTO of Earendil, which builds the open-source model harness Pi. âÂÂThe user has to say, okay, if this only comes back with 50% probability, maybe this is a coin toss, and I disregard it. But if itâÂÂs 95%, sure, then I can do something with it.âÂÂ
Another potential use for Jev is model routing, Ronacher said. Predicting whether a given workload requires a specific model would be useful, but using an LLM for the job would be expensive. JevâÂÂs low cost and speed make that kind of real-time sorting possible.
And thatâÂÂs AlmeidaâÂÂs hope. The model is named after William Stanley Jevons, the 19th-century economist whose eponymous paradox describes how the falling cost of a commodity can lead to it being used more and more. In this case, the falling cost of intelligence should lead to its widespread deployment.
âÂÂWe think that thereâÂÂs just going to be smart software all over the place in a way thatâÂÂs emergent and distributed ⦠much more like the early internet than you know like the the mega apps that people are trying to build right now,â Almeida said.
Almeida is tight-lipped about the modelâÂÂs architecture, which outside observers suspect is built on top of an open-weight LLM. The company refers to Jev as a âÂÂSystem One model,â focused on intuition rather than reasoning, and specifically focused on the right task. Almeida says Jev is trained exclusively on synthetic data using a technique he calls âÂÂreinforcement learning from calibrated decisions.âÂÂ
âÂÂWe made an early bet that we will be making all of our data, and that has been one of the best bets IâÂÂve ever made in my lifeâÂÂbetter than our launch, in my opinion, better than RLHF,â he told TechCrunch. âÂÂHalf of [our company] is a lab that basically owns this entire subfield of statistically well-understood synthetic data, and that is now my life joy.âÂÂ
For now, Jev stands alone as this kind of model, but Ronacher expects that competitors will spring up now that its utility is apparent.
âÂÂWe should have seen this earlier in many ways, but presumably because the LLMs are so cheap and subsidized, you often donâÂÂt have to be creative yet,â he said.
TypeSafe itself will be building more versions of the model, in new modalities. Asked if TypeSafe is a frontier lab, Almeida said, âÂÂthe main product of Frontier Labs is fear or hype. I would like our main product to be intelligenceâ¦[but we are] not a lab in the sense of, you know, like bet on infinite wealth, or a religion, or building God in a data center, or whatever is the the thing of today.âÂÂ
Topics
When you purchase through links in our articles, we may earn a small commission. This doesnâÂÂt affect our editorial independence.
