This AI Model Does’t Use Words?!
Typesafe AI released Jev, a new AI model that outputs structured decisions (choices, scores, or booleans) instead of generated text like traditional LLMs. This approach makes Jev dramatically cheaper and faster, with input costs at 4 cents per million tokens and free output tokens.
Summary
A new AI model called Jev from Typesafe AI is generating significant interest due to its fundamentally different approach compared to models like GPT. While traditional language models generate new text and code to make decisions, Jev skips the text generation step entirely and outputs structured values directly. The model supports three types of structured outputs: choices, scores, and binary values (true/false). This architectural difference provides substantial practical advantages. The cost efficiency is remarkable—input tokens cost only 4 cents per million tokens (or $42 per billion tokens), while output tokens are free because they are too inexpensive to meter separately. Beyond cost, Jev offers significant speed advantages over traditional LLMs since it doesn't need to generate lengthy text outputs. To demonstrate the model's capabilities, Steven Tay created an application that uses Jev to detect whether URLs are malicious before users visit them. The implementation is straightforward; developers can use coding assistants like Codeex or Claude Code to specify what they want Jev to do, and these tools can generate the necessary code to run Jev.
Key Insights
- Jev differs from GPT-style models by skipping text generation entirely and outputting structured values directly for decision-making
- Jev supports three types of structured outputs: choices, scores, and boolean values (true/false)
- Jev's input tokens cost 4 cents per million tokens while output tokens are completely free because they are too cheap to meter
- Steven Tay demonstrated Jev's practical capability by building an app that detects malicious URLs before users visit them
- Developers can easily create Jev applications by using coding assistants like Codeex or Claude Code to generate the implementation code
Topics
Transcript
[0:00] But there was something else causing a lot of buzz this week. And that's a model out of a company called Typesafe AI called Jev. When you think of like the GPT models, those create new text. Even when you ask it to make decisions for you, it's still creating text and writing code that then goes and makes those decisions. Jev, on the other hand, just skips straight to the decisionmaking. The output from existing LLM are, you know, strings and generated text. The outputs from Jev are structured values. And there's really [0:30] three types of structured outputs. You can give it a choice, a score, or a new, which is basically, you know, true or false.…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Matt Wolfe
AI News: Dots, GPT-6.1 Sol, Sonnet 5.5, Gemini 4, and everything you need to know
A comprehensive review of major AI announcements from the week, including OpenAI's new Dots agent, GPT-6.1 Soul model, Anthropic's Sonnet 5.5, and Google's Gemini 4 Argon, with analysis of pricing, capabilities, and competitive positioning across different models.
The Hands Down Best Coding Model Right Now
The speaker reviews Opus 5.5, claiming it's currently the best state-of-the-art coding model at a reasonable price point. They demonstrate its capabilities by running a 20-hour test where the AI built a nearly complete recreation of the Mega Bonk game, including accurate character tiers, sound design, and UI elements.
AI News: Opus 5.5, GPT-6 Sol, Jev, Muse and More!
This week in AI saw major announcements from Meta Connect (Muse agent, new AR/VR glasses), three new language models (Claude Opus 5.5, GPT-6 Soul, Grok 4.7), significant buzz around Typesafe AI's Jev decision model, and new features from YouTube, Microsoft, Google, and Spotify. Claude Opus 5.5 emerged as the new state-of-the-art model while Jev introduced a fundamentally different approach to AI outputs focused on decision-making rather than text generation.
5 Ways To Use ChatGPT's New Features
The transcript demonstrates five practical applications of ChatGPT's new features, including computer use for software automation, advanced image editing, cloud browser integration for account access, voice-powered live agents, and persistent session agents for long-term user interaction tracking.
My BIGGEST AI Pet Peeve...
The speaker criticizes companies for announcing features before they're ready, using Claude's Co-work announcement as an example. They argue that companies should only announce features once they're fully available, not during rollout phases.