Jev by Typesafe AI

What hackers pick instead

• LayaOpen-weight, self-hostable decision model that people run locally and fine-tune for specific classification tasks, with fast responses. Common caveat: weak out-of-box accuracy and a 1024-token context limit.44recommend
• Cloudflare ClefOpen-weight decision model people like for its open-source license, self-hosting and image input; critics say it is slower, costlier and less accurate than Jev.21recommend
• OpenJevFast, cheap structured-output model that cannot emit tokens outside a given set. Critics doubt it differs from other small LLMs, and the naming is confusing.21recommend
• Typesafe AIPeople recommend it as a cheap API model that handles new classification problems without fine-tuning. Critics say it is over-hyped and not production-ready without fine-tuning.16recommend
• Strands Decider 2BA small open-source decision model that people run locally for filtering and automations. Critics note a bias toward the first option listed.7recommend

Every quote

The 40 most detailed of 716 comments. Each links to the original.

1. the real win with jev isnt beating a trained classifier on accuracy, its that you can change what you're classifying by editing a prompt instead of retraining and redeploying a model.

   Recommends · Sep 2026 · in “OpenAI is well positioned to fast-follow Jev”

2. the idea that you can get a general classifier that is very accurate across any domain and on any modality with minimal latency and a very low price point is wild

   Recommends · Sep 2026 · in “Ollaya – Ollama for open-source, Jev-style decision models”

3. I've been running jev in some specific use cases and see much better speed and better accuracy than just using structured output from gemma/qwen.

   Recommends · Oct 2026 · in “Jev meets the Trolley Problem”

4. Especially the “zero hallucination” output and the (if it works how the documentation make it sound) prompt injection resistant architecture.

   Recommends · Oct 2026 · in “Typesafe AI raises $870M at $7.5B”

5. The largest change is the packaging; An api with a tolken based pricing, and a schema to define the output structure for quick setup.

   Mixed · Sep 2026 · in “OpenAI is well positioned to fast-follow Jev”

6. I agree it's overhyped, but the transition to a general purpose classifier (vs a narrow scope classifier) is new and noteworthy.

   Mixed · Sep 2026 · in “Show HN: Jev Plays Pokémon Red”

7. There is basically no reason to trust someone at their word about data retention when you can run it yourself as easily as this.

   Recommends · Oct 2026 · in “We’re releasing Jev 1.14.0 along with experimental features for dreamers”

8. Jev is much better than the traditional ML crowd gives it credit for, but my enthusiasm hits a wall when it comes to their data

   Mixed · Sep 2026 · in “Kev: Tiny Jev-like family of decision models built on top of Qwen3.5”

9. Jev gets you the zero-shot, zero-infra benefits of LLMs, while being closer to the speed and cost of traditional ML classifiers

   Recommends · Sep 2026 · in “OpenAI is well positioned to fast-follow Jev”

10. It was specifically the latency/price point Jev offered vs. the general accuracy it claimed that generated all the excitement.

    Mixed · Sep 2026 · in “Ollaya – Ollama for open-source, Jev-style decision models”

11. You’re correct, it’s a great solution for a very narrow set of high volume classification needs that require very low latency.

    Mixed · Oct 2026 · in “Typesafe AI raises $870M at $7.5B”

12. Small, fast, essentially trivial to locally run - this alone makes them an actual candidate for real-world planning systems.

    Recommends · Oct 2026 · in “Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers”

13. The core feature is good quality out-of-the-box domain and cross-domain calibrated probability you can use as a threshold.

    Recommends · Sep 2026 · in “Jev Based Code Review”

14. decision models/jev - are generic, you can throw them at most generic classification problems, and they are good enough.

    Recommends · Oct 2026 · in “Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers”

15. Jev does perform at a more consistent level, allegedly roughly akin to GPT-5.6 Terra, but at the lower cost and latency

    Recommends · Sep 2026 · in “OpenJev”

16. It is impressive, but all the hype and fake demos are selling it as a model that is as smart as frontier reasoning LLMs

    Mixed · Sep 2026 · in “Show HN: Jev Plays Pokémon Red”

17. jev (specifically) is so dirt cheap and fast that you can throw it at things that should not be justifiable in the past

    Recommends · Oct 2026 · in “Build your own decision model”

18. it allows you to define a new semantic decision at runtime in natural language, without training a model for that task

    Recommends · Sep 2026 · in “I was impressed by Jev, please explain why I shouldn't be”

19. The consistency was nice to see, BUT, as someone who trained NLP models prior to LLMs, it's just BERT with more data.

    Mixed · Sep 2026 · in “I built non-autoregressive decision models with RL a year ago”

20. my conclusion was that the speed and the confidence score come from reading out a single token, not from Jev itself

    Mixed · Oct 2026 · in “Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers”

21. Jev as a concept will always provide more reliable decisions / outputs than LLM-as-a-judge giving a scalar score.

    Recommends · Oct 2026 · in “Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers”

22. the huge benefit in real systems for Jev like solutions i see is the cybersecurity / prompt injection mitigation

    Mixed · Sep 2026 · in “I built non-autoregressive decision models with RL a year ago”

23. On actual NLP problems (including ones with uncertainty under human review) it does appear to be well calibrated

    Mixed · Oct 2026 · in “How accurately calibrated is Jev?”

24. If you ask Jev similar input + questions, it's pretty stable. And does a reasonable job on a lot of questions.

    Recommends · Sep 2026 · in “Kev: Tiny Jev-like family of decision models built on top of Qwen3.5”

25. Jev is prompted with natural language, so it is flexible and a good fit to replace subagents for certain tasks

    Recommends · Sep 2026 · in “OpenAI is well positioned to fast-follow Jev”

26. My AI harness uses Jev or a compatible model to approve shell commands that look safe without asking the user.

    Recommends · Oct 2026 · in “Clef: Open-weight decision models, and new RL fine-tuning platform”

27. What really pissed me off was that y'all got manipulated into thinking it was something revolutionary, while really it was a marketing stunt.

    Criticizes · Oct 2026 · in “Jev meets the Trolley Problem”

28. There really isn't any architectural magic to Jev, it's just a straightforward application of normal LLM tech with some good marketing.

    Criticizes · Sep 2026 · in “Jeff – Jev-compatible 0.8B decision models, trained at home, ~30 ms”

29. Jev's "confidence" output is fake/redundant - it's just a formula applied to probabilities, it conveys no additional information.

    Criticizes · Sep 2026 · in “OpenAI is well positioned to fast-follow Jev”

30. They know it's not. The company actually has or had public statements that they didn't like public benchmarks for comparison.

    Criticizes · Sep 2026 · in “Show HN: Jev Plays Pokémon Red”

31. specifically calling this kind of model a breakthrough, without crediting previous art, and keeping everything closed source

    Criticizes · Sep 2026 · in “I built non-autoregressive decision models with RL a year ago”

32. None of that makes sense, if you can get 90% of the performance from an open source model that's been available for years.

    Criticizes · Sep 2026 · in “I built non-autoregressive decision models with RL a year ago”

33. it was 0-for-3 at giving yes/no probabilities that were particularly close to the obviously correct no-information answers

    Criticizes · Sep 2026 · in “Jev and System One Models: Calibration Beats Accuracy”

34. every single demo I saw was either fake (e.g. playing games), contrived, or plain wrong (e.g. using Jev for compaction)

    Criticizes · Oct 2026 · in “Typesafe AI raises $870M at $7.5B”

35. I don't see what's so exciting about it, compared to regular LLMs which support structured output options in the API.

    Criticizes · Oct 2026 · in “Typesafe AI raises $870M at $7.5B”

36. This one public benchmark (the only I've seen) seems to give the open versions way too much credit.

    Criticizes · Sep 2026 · in “Kev: Tiny Jev-like family of decision models built on top of Qwen3.5”

37. They do present evidence that Jev may not be the best choice for these two particular scenarios, but then they draw broad conclusions from their tests that are not justified.

    Mixed · Oct 2026 · in “Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers”

38. Then everyone breathlessly repeats this story about how Jev is useless because open source models they never tried claim to do the same thing and better.

    Recommends · Oct 2026 · in “Typesafe AI raises $870M at $7.5B”

39. Fortunately Jev is cheap, so I dont think it matters too much, but I think its robbing people of the opportunity to learn and implement this themselves.

    Mixed · Oct 2026 · in “Typesafe AI raises $870M at $7.5B”

40. Jev is incredibly powerful for the decision making part of it when presented with a scenario that gives it a little context

    Recommends · Sep 2026 · in “Show HN: Jev Plays Pokémon Red”