Open-source Jev clones proliferate weeks after TypeSafe's launch
Multiple implementations of decision models flood GitHub as developers race to replicate TypeSafe's fast, single-pass classification approach.
What to know
- At least five open-source decision-model implementations appeared within four days (Ollaya, TinyJev, Valen, Nanojev, GLM-5.3-Flash variant), all claiming to replicate or improve on TypeSafe's Jev.
- The rapid copying has sparked debate: some dismiss decision models as trivial wrappers; others defend them as genuinely novel; most agree the real moat lies in the RL training data pipeline, not architecture.
- Ollaya emerged as the most polished alternative with API compatibility, multiple model variants, and sub-20ms latency, potentially accelerating TypeSafe's commoditization.
- Unresolved: whether Laya existed before Jev's public launch, whether decision models are materially different from re-rankers, and whether TypeSafe's competitive advantage survives perpetual open-source parity.
The dispute Is Jev's innovation a durable competitive advantage (requiring the training data/RL pipeline to replicate), or are the architecture and scripts trivial, making open-source parity a death knell for TypeSafe's business model? · positions read across 67 posts and comments
Decision models are genuinely novel and valuable despite rapid OSS copying; the training pipeline and ongoing innovation are the durable moat.
-
“For everyone dismissing Jev's innovation as being trivial, no it's not. It is definitely not the MNIST classifier you had trained in 2019.”
fooker · Hacker News ↗
The core architecture is trivial to copy; OSS replication within weeks proves the innovation isn't defensible and questions whether TypeSafe has a sustainable business.
-
“I'm not sure what this means for AI startups if their innovations can be copied by OSS so quickly (what, like 2 weeks?). There's "consumer surplus" for everyone, to borrow an economic concept. But we do ideally want some of the surplus to…”
pradn · Hacker News ↗
Unclear what decision models solve; the use cases and advantages over existing techniques (re-rankers, structured output, fine-tuning) are not compellingly demonstrated.
-
“I installed it, I tried the examples, it works.... But forgive my lack of imagination... what is this useful for?”
solaire_oa · Hacker News ↗
Performance quality varies; Laya (and other OSS variants) underperform Jev on real workloads, suggesting the originator's edge is still meaningful.
-
“Has anyone actually seen better or the same results with Laya compared to Jev? From my experience, Laya performs significantly worse. It's less confident and often makes wrong decisions with more complex queries.”
george_max · Hacker News ↗
TypeSafe Creator of Jev decision modelOllaya maintainers Open-source implementation leadsTinyJev maintainers Open-source implementation leadsValen authors Multimodal extension developers
How it unfolded 9 developments, newest first · click a bar or a number to jump postscomments
-
9
Adaptation of GLM-5.3-Flash published as Jev-style decision model
Another OSS variant emerges, this time adapting Alibaba's GLM-5.3-Flash into a System One decision model via PrivateMode's blog.
“I can't wait for a version of this to come along that supports images. I want to build a feature into Digital Carrot for creating AI goals where you can create a daily goal to, for example, "empty the dishwasher" that you would then verify by taking a picture of the empty dishwasher at the end of the day.”
— newswangerd, Hacker News commenter · source -
first by HN Frontpage, 1d ago
-
I imagine it’s not so hard to optimize a model for this use case.Off the top of my head, I would skip all the modern linear attention / state space stuff and use classical attention. But run prefill in a fully sliding-window mode so that “state” tokens simply don’t attend to far away tokens, or maybe also allow everything to attend to the first…
2 more of the top 3 · 17 posts in this stretch
-
I have run tests with qwen 3.8 and gemma 4 in a way similar to this post (based on an open source project that also does this with gemma4).Getting competitive accuracy with Jev is fairly easy, if by accuracy you mean that the highest weighted answer is the right one. GLM 5.3 is complete overkill, much smaller llms will doWhat Jev brings to the…
-
20k tok/sec prefill on B200/B300 isn't particularly noteworthy for medium-sized models like GLM-5.3-Flash, vLLM and SGLang achieve it on a reasonable number of models, especially at NVFP4.50k tok/sec is pretty impressive though.But... when you were doing your measurements, were you using the same random book excerpt? If you were potentially…
-
-
8
Commenter questions whether Laya predates Jev announcement
A user raises the possibility that Laya (or its predecessor) shipped before or independently of Jev, challenging the narrative that OSS copied the commercial product in two weeks.
-
> I'm not sure what this means for AI startups if their innovations can be copied by OSS so quickly (what, like 2 weeks?).My thoughts too. It sounds like a minor feature being framed as a whole new business.Then again, Dropbox and Docker are too.
1 more of the top 2 · 3 posts in this stretch
-
Didn't llaya come first? So the entire Jev's innovation is encapsulating idea in a cheap service?
-
-
7
Observers pinpoint training pipeline, not architecture, as real moat
Commenters argue that the RL synthetic data pipeline and training methodology—not the model architecture—represents the defensible innovation, and that open-sourcing those would be the true competitive threat.
“The moat is the RL synthetic data pipeline they set up to train jev. Open sourcing that would be the coup, not the model architecture and training scripts, which are trivial.”
— _menelaus -
I think the idea of a "feature startup" is dead. What used to be a niche subscription business is now an individual Epic level of work. The smallest viable business becomes what two or three years ago was a mid tier enterprise. It is no longer "look at this tool I maintain", but "we take this specific approach using these hundreds of tools merged…
2 more of the top 3 · 6 posts in this stretch
-
Today we might need to evaluate “innovation” in a new standard, and have a different expectation for what innovator would be awarded. Getting public attention in such an era where innovation happens every a few days could’ve already been something precious. And that attention would allow TypeSafe to be heard easily next time. Like OpenAI…
-
For everyone dismissing Jev's innovation as being trivial, no it's not.It is definitely not the MNIST classifier you had trained in 2019.The difference is that you only train it once and the modern LLM machinery sort of takes care of that with large contexts.It's great that Jev proved this is a viable product. I'd expect a great many research…
-
-
6
Analysts warn rapid open-source copying threatens startup defensibility
Community members note that OSS implementations replicated the core innovation within two weeks, questioning what moat TypeSafe retains and whether the business model survives if competitors copy perpetually.
“I'm not sure what this means for AI startups if their innovations can be copied by OSS so quickly (what, like 2 weeks?). There's "consumer surplus" for everyone, to borrow an economic concept. But we do ideally want some of the surplus to flow to the innovator, too.”
— pradn -
I'm not sure what this means for AI startups if their innovations can be copied by OSS so quickly (what, like 2 weeks?). There's "consumer surplus" for everyone, to borrow an economic concept. But we do ideally want some of the surplus to flow to the innovator, too. I know there were precursors, but that's fine - it's hard to have a totally novel…
2 more of the top 3 · 8 posts in this stretch
-
Because what they did is kinda trivial. Its basically like the Dropbox comment really[0], except here you don't need petabytes of storage and infinite VC pockets.After chatgpt everything in AI mostly became LLMs and building wrappers around them. It's like people forgot how to do ML.To those of us who actually trained models back in the day, its…
-
I installed it, I tried the examples, it works.... But forgive my lack of imagination... what is this useful for?Like, their example is of classification for a support interface.... `refund_requested`. Pretty convenient bool given the example is about a refund- what if 99% of submissions don't ask about a refund? Also, is that user not a…
-
-
5
Community questions distinguish decision models from existing re-rankers
Developers debate whether decision models are genuinely novel or simply re-branded re-rankers with better probability calibration, with questions raised about what differentiates them architecturally.
-
Guys I have a real q, what is the difference between an instruct based re-ranker and laya/jev I just don't see it.Edit: One is that jev/laya are tuned to have better probabilities, but a reranker can be fine tuned to do that as well. And jev/laya use RLCD?
1 more of the top 2 · 2 posts in this stretch
-
It would be really cool to have LLMs and System One in a single tool - in this case, if Ollama implemented it.
-
-
4
Ollaya launches as full-featured local decision-model framework
The most polished open-source implementation arrives with four model variants (laya, decider, von, qwen3guard), desktop and CLI tools, Docker support, and API compatibility with TypeSafe's SDKs. Achieves 10ms latency on RTX 4090 for five-question requests.
“Ask typed questions about any text or JSON and get calibrated answers in milliseconds. Private, open source, on your own hardware.”
— Ollaya documentation -
Has anyone actually seen better or the same results with Laya compared to Jev? From my experience, Laya performs significantly worse. It's less confident and often makes wrong decisions with more complex queries.
2 more of the top 3 · 3 posts in this stretch
-
Cool... but this does seem undermined by the fact that Ollama can add support for decision models at any time.
-
>Run decision models locally.>example is a text classification task instead of a decision
-
-
3
Valen extends decision models to images and video
A multimodal variant emerges, adding visual perception to the System One decision pattern with Qwen3.5-0.8B and 2B backbones, solving visual puzzles in ~9 decisions with 1.13 seconds latency.
- 1 day quiet
-
2
TinyJev launches with two model sizes, offline-capable
TinyJev released with 0.6B and 4B parameter sizes, runs entirely offline on MLX (Apple) and PyTorch, achieving 86ms per question on base M1 versus GPT-6 Sol's 2,042ms on the same benchmarks.
“TinyJev scores every option of every question in a single forward pass and returns all twelve together, 596 ms, 10 of 12 right. GPT-6 Sol writes the same twelve as JSON: 2,198 ms, 12 of 12 right.”
— TinyJev documentation, Project benchmark · source -
1
Nanojev released as minimal 200-line Jev clone
A stripped-down, single-file implementation of decision models appears on GitHub, demonstrating the concept can be reproduced in minimal code.
What people are saying 14 voices from 1 site · best of 67 · verbatim
- Does Laya predate Jev's public announcement, or was it genuinely developed in response?
- How materially different is a decision model from a fine-tuned re-ranker or classifier with probability calibration?
- What happens to TypeSafe's business if Ollaya or another OSS tool achieves parity on latency and accuracy?
- Today
-
It seems people are just guessing at the architecture behind Jev. Obviously the functionality itself is easy to replicate, but why Jev seems to be making such a splash (beyond the doh! factor of it's huge applicability) is the ultra-low cost and speed, which may be due to architecture.The Laya model compared in TFA shows one way Jev may be getting…
-
You can turn any sufficiently smart LLM into yes/no decision model or equivalent. I already have an existing workflow with a two paragraph detailed prompt, that sends pages of stuff to an LLM and asks it to return only 7 JSON objects. Several of those objects are binary "yes or no" choices of like, whether the content contains certain things.You…
- Yesterday
-
There is also a small wrapper over an unmodified llamacpp which implements something like this (raw label softmax, not calibrated!) on any llamacpp supported model. It is different than Leya/Kime here but can be useful for someone too.
- Sep 25
-
It would be good to list 1) zero-shot accuracy and 2) latency on the models page . The LLM-based models' latency is probably much higher than the BERT approaches I would assume.Also curious, it seems from looking at the accuracy scores you gave that it seems to be NLI > Gliclass > Laya (for Bert types)? Why do you seem to feature/recommend Laya…
- Sep 21
-
Now that the models are smarter / know more than most of us, a new bottleneck is finding out what you don't know. If you don't know about logistic classifiers and how they could be applied to your problem, you will not ask for it.These days, I tend to start my coding sessions by the high level problem I'm trying to solve vs the prescriptive…
-
I'm working on exactly this! I'm building a small model that classifies the correct DOM node containing an HTML's article content/title/date/author (given a raw html with a lot of noise/chrome). A fun learning exercise :)30KB model, 40-50ms inference. Pretty happy with the results so far!I can see an entire industry of tiny models like this, now…
-
I wrote a small proxy that points points to a jev api and a frontier api. My harness connects to the proxy and only sees the frontier api, models, commands, etc. When I send prompts with tool calls, proxy routes to jev, jev narrows the tools, proxy cleans/sends to the frontier api.So far in my tests, about 60% less tool calls. I'm also going to…
-
I built a whole thing that collects data, trains classifiers, exports models and dataset just for that. Claude writes me a terraform file that contains shape of the classifier and dataset. For images it can create datasets based of another dataset (crop this region from images that have these labels).Originally it was so I can label data to…
-
Do you realize people is using LLM's as classifiers, right? For lots of companies and developers reaching an API is feasible, while running a training pipeline, no matter how simple, is not. I know that they should still be gathering data for evaluation and they could use it to train a model instead. But they won't do it, for a variety of…
-
If you only need classification, and you can provide some training data, you can ask Codex/Claude to build an embeddings + logistic classifier model for youFor emails, I get 95% accuracy with this method, with only 50-100 examples for trainingTraining the model takes less than 5 minutes on a CPUThe resulting model is <1MB, and inference is sub…
-
Man, I'm already burnt out on all this jev talk.The one thing jev has going for it is a dedicated company focused entirely on making the product good and keeping it maintained. I haven't been willing to jump on board with all these jev-shaped projects because their releases feel driven mostly by opportunism. I'm fine waiting a bit for the…
-
The open source ones- I downloaded a number and tried them and compared to Jev.Anything that required knowledge / familiarity mmBERT and ModernBERT post-trains performed much worse.So it seems like they did some kind of useful expansive pre-training.Things that were Qwen or Gemma Diffusion did better at those kinds of tasks but were generally…
-
On Gemma 4 12B, I am getting 220 ms per move or QS. I used it to play the Snake game locally:prompt_eval=244 ms wall=245 ms schema_cache=hit generated=0Move limit reached after 200 moves: score=16, length=19.So, if a 12B dense model can offer this latency on a local old PC, then definitely you can scale it up with more powerful machines and get…
-
I think a great use case for these will be when they have large context windows and are able to enforce styling rules for frontend development, and component creation rules for react. You can then ditch the styles guides and styling skills and create a decision tree for enforcing styling, so that you can't run into drift issues or duplication…