正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

What is a decision model? Jev by TypeSafe AI answers yes/no and multiple-choice questions, with a confidence score. Much of software development are a sequence of decisions, and Jev is 10x cheaper and faster than an LLM. Let’s understand this through practical examples:

OpenRouter

146,557 subscribers

150,351 次观看 • 2 天前 •via X (Twitter)

教育科学技术

Anya Rossi• Live Now

Private livecam show

27 条评论

OpenRouter 的头像

OpenRouter2 天前

1/ Better support triage. Ask it to tag the tickets: cancelling? Refund? Angry? Bug report? Escalate? Jev tagged the messages in 2.3 seconds for a tenth of a cent. A chat LLM on the same batch took 20 seconds.

OpenRouter 的头像

OpenRouter2 天前

2/ Faster agent tool calls. Agents need to make tool calls, Jev could pick the right tools to call at a fraction of the cost and time.

OpenRouter 的头像

OpenRouter2 天前

3/ Quickly filter an X feed. Every reply on X is a multiple-choice question: genuine question, hot take, promotion, rage bait, or troll? Filter as you scroll, in real-time, for pennies.

OpenRouter 的头像

OpenRouter2 天前

4/ Real-time actions in meeting notes. Per conversation - was it addressed to the assistant? Did the person finish their thought? Was there a specific request? Was there an action to take? Done real-time.

OpenRouter 的头像

OpenRouter2 天前

5/ Break problems into small yes/no and multiple-choice decisions, and a decision model answers them fast enough and cheap enough to run on every message, every tool call, every sentence.

OpenRouter 的头像

OpenRouter2 天前

6/ All mini apps are in OpenRouter Labs, with the code included: Model page:

EKOS _ AGI 🦊 🇮🇷 的头像

EKOS _ AGI 🦊 🇮🇷2 天前

@typesafeai

lewington 的头像

lewington2 天前

@typesafeai yeah a lot of software is basically just making little decisions all day. using a full llm for every one of those does feel kinda unnecessary

StrangerNerd 的头像

StrangerNerd1 天前

@typesafeai I want a open weights of this kind of model

GTheMaximalist 的头像

GTheMaximalist1 天前

@typesafeai I'm running my own version of jev locally, it's getting response faster than JEV at 0 cost and it's open source, you don't have to send any data through the API

Justin Smith 的头像

Justin Smith1 天前

@typesafeai I can’t get over your choice in background when you’re on Mac! I can’t remember what Windows OS that was from. Was it XP?

EKOS _ AGI 🦊 🇮🇷 的头像

EKOS _ AGI 🦊 🇮🇷2 天前

@typesafeai Making decisions 10x cheaper and faster is powerful. The next piece is provenance: can we trace each decision back to the exact knowledge, evidence and context that produced it?

Aapakari 的头像

Aapakari1 天前

@typesafeai Set a confidence cutoff so weak calls escalate to humans

First Sauce Labs 的头像

First Sauce Labs1 天前

@typesafeai The "10x cheaper and faster" claim on a yes/no is always going to get my attention. Sounds like a good fit for constrained classification.

FoundationBuilder 的头像

FoundationBuilder1 天前

@typesafeai Very nice to have Jev on OpenRouter now!

Robot Notes 的头像

Robot Notes1 天前

@typesafeai vercel just posted adoption numbers: ~13% of AI Gateway teams used Jev on day one. 2x the GPT-5.6 family, 6x Fable 5.1

peyCyber 的头像

peyCyber1 天前

@typesafeai how are you benchmarking the confidence scores against llm baselines

Aarav Dayal 的头像

Aarav Dayal1 天前

@typesafeai Is this @pingToven

RAZA | AI EXPLORER 的头像

RAZA | AI EXPLORER1 天前

@typesafeai Interesting approach. Specializing in decisions rather than text generation could make AI workflows faster and more efficient for many practical tasks.

Fajar M Reza 的头像

Fajar M Reza1 天前

@typesafeai Decision models shine when bounded outputs cut latency, cost, and verification burden.

ShadowAguy 的头像

ShadowAguy1 天前

@typesafeai 10x cheaper and faster" than what exactly? A bigger LLM doing the same thing, or a bigger LLM doing everything else too?

The Chieftain A 的头像

The Chieftain A1 天前

@typesafeai So if we have copilot in our org, can jev use that language model for decisions?

Tony 的头像

Tony2 天前

@typesafeai Paying nothing on output because the answer is a typed choice is a neat consequence. Routing and gating is where I'd put it first 👍

AI Mastery Guide 的头像

AI Mastery Guide1 天前

@typesafeai decision models actually make sense as their own category

Ryan 的头像

Ryan1 天前

@typesafeai the confidence score matters more than the label. a cheap model that knows when its option set doesn't fit can route the weird cases upward instead of silently forcing a bad branch.

BayesChat 的头像

BayesChat1 天前

@typesafeai The support-triage example is a good one. Fast, cheap classification seems like a much better fit than using a full chat model for every small decision.

James Malsawm 的头像

James Malsawm1 天前

@typesafeai What's the limitations of Jev?

相关视频

Jev Engineering is the layer most agent stacks are still missing. state → decision → action → verification → next state up to 193x faster and 444x cheaper in our tests. everyone is optimizing the model. Jev Engineering optimizes what happens between model calls. every agent eventually hits the same fork: state → multiple possible actions → one route has to win instead of sending every decision back to an expensive LLM call, Jev turns that into a dedicated layer: so the loop becomes: LLM → reason Jev → decide tools / agents → execute state → update Jev → decide again once decisions become their own engineering layer, you can benchmark, batch and verify them separately. full blueprint in article ↓

Jev Engineering is the layer most agent stacks are still missing. state → decision → action → verification → next state up to 193x faster and 444x cheaper in our tests. everyone is optimizing the model. Jev Engineering optimizes what happens between model calls. every agent eventually hits the same fork: state → multiple possible actions → one route has to win instead of sending every decision back to an expensive LLM call, Jev turns that into a dedicated layer: so the loop becomes: LLM → reason Jev → decide tools / agents → execute state → update Jev → decide again once decisions become their own engineering layer, you can benchmark, batch and verify them separately. full blueprint in article ↓

Ricker

81,664 次观看 • 1 天前

Jev by TypeSafe AI is now on the Venice API, in beta. It answers, it doesn't write. Send it your app's state and typed questions. Get back a typed answer your code can branch on: a probability, a chosen option, or a score on your rubric. No JSON to coax out of a chat model.

Jev by TypeSafe AI is now on the Venice API, in beta. It answers, it doesn't write. Send it your app's state and typed questions. Get back a typed answer your code can branch on: a probability, a chosen option, or a score on your rubric. No JSON to coax out of a chat model.

Venice

471,014 次观看 • 1 天前

Jev v1.13 dodges rockets with probability calculation 🚀 TypeSafe AI's new non-LLM model returns decisions instead of text so we had it calculate a safe tile every 330 ms while rockets fell and it survived 25 of 26 for under a cent Run Jev via API ->

Jev v1.13 dodges rockets with probability calculation 🚀 TypeSafe AI's new non-LLM model returns decisions instead of text so we had it calculate a safe tile every 330 ms while rockets fell and it survived 25 of 26 for under a cent Run Jev via API ->

atomic.chat

54,019 次观看 • 3 天前

Jev Founder, Diogo Almeida (ex-OpenAI): "The next era is not the Claude Code era. Claude Code is still part of the assistance era - Jev is the next step Jev is x200 faster, x400 cheaper and has 0 hallucination. This is what comes after RLHF" In 17 minutes he explains why every model you use has a human-shaped flaw baked into it and why Jev is better worth more than a $500 Claude Code engineering course watch today, then read how to actually boost yourself x100 with Jev in the post below

Jev Founder, Diogo Almeida (ex-OpenAI): "The next era is not the Claude Code era. Claude Code is still part of the assistance era - Jev is the next step Jev is x200 faster, x400 cheaper and has 0 hallucination. This is what comes after RLHF" In 17 minutes he explains why every model you use has a human-shaped flaw baked into it and why Jev is better worth more than a $500 Claude Code engineering course watch today, then read how to actually boost yourself x100 with Jev in the post below

Codez

175,262 次观看 • 2 天前

TypeSafe AI released Jev. I truly believe this is a game changing kind of AI model. No joke. It does NOT replace LLMs. But in tandem with LLMs it feels like a missing building block. It makes the kind of applications you can build now SO much better and more powerful. And cheaper! It's an AI model that's all about decision making. But not like old-school classifiers ("fraud or no fraud") but with flexibility you know from LLMs. Feed some input (eg customer email, any other text, structured data) into it, define your questions ("which team should answer this email?") and dimensions ("sales, technical, fraud") and get back a response within milliseconds. No long reasoning, no exploding token costs. Sure, that answer will still be generated by an LLM, complex questions may still require an (additional) LLM but this model is really mindblowing.

TypeSafe AI released Jev. I truly believe this is a game changing kind of AI model. No joke. It does NOT replace LLMs. But in tandem with LLMs it feels like a missing building block. It makes the kind of applications you can build now SO much better and more powerful. And cheaper! It's an AI model that's all about decision making. But not like old-school classifiers ("fraud or no fraud") but with flexibility you know from LLMs. Feed some input (eg customer email, any other text, structured data) into it, define your questions ("which team should answer this email?") and dimensions ("sales, technical, fraud") and get back a response within milliseconds. No long reasoning, no exploding token costs. Sure, that answer will still be generated by an LLM, complex questions may still require an (additional) LLM but this model is really mindblowing.

Maximilian

36,117 次观看 • 3 天前

$Now we're cooking. Got TypeSafe AI's Jev + AXe controlling the iOS simulator now ultra fast at a fraction of the cost of using an LLM. This is game changing!$

Now we're cooking. Got TypeSafe AI's Jev + AXe controlling the iOS simulator now ultra fast at a fraction of the cost of using an LLM. This is game changing!

camsoft2000

103,967 次观看 • 3 天前

Jev Founder, Diogo Almeida (ex-OpenAI): "The next era is not the Claude Code or Codex era, they are still part of the assistance era with human in the loop - JEV is what comes next for LLMs x200 faster, x400 cheaper, 0 hallucination, no human in the loop - that's JEV, this is how LLMs will look like" in 36-minute tech talk, Jev Founder explained why RLHF isn't a thing anymore and how modern LLMs will be built this talk is worth more than a Stanford Machine Learning degree watch today no matter what, then learn how to become a Jev Engineer in the article below

Jev Founder, Diogo Almeida (ex-OpenAI): "The next era is not the Claude Code or Codex era, they are still part of the assistance era with human in the loop - JEV is what comes next for LLMs x200 faster, x400 cheaper, 0 hallucination, no human in the loop - that's JEV, this is how LLMs will look like" in 36-minute tech talk, Jev Founder explained why RLHF isn't a thing anymore and how modern LLMs will be built this talk is worth more than a Stanford Machine Learning degree watch today no matter what, then learn how to become a Jev Engineer in the article below

Codez

215,589 次观看 • 1 天前

Also have been playing with TypeSafe AI Jev, insane! So many immediate use cases and new apps are possible. What a time to be a builder! Sharing some experiments here starting with: Keystroke oracle / predictive launcher: Your launcher ranks by aliases, fuzzy match, and habit. Jev reads intent: type "the pdf I just downloaded" and the newest PDF is already the top hit with a full confidence on every keystroke, in ~100 ms

Also have been playing with TypeSafe AI Jev, insane! So many immediate use cases and new apps are possible. What a time to be a builder! Sharing some experiments here starting with: Keystroke oracle / predictive launcher: Your launcher ranks by aliases, fuzzy match, and habit. Jev reads intent: type "the pdf I just downloaded" and the newest PDF is already the top hit with a full confidence on every keystroke, in ~100 ms

nader dabit

418,779 次观看 • 2 天前

does jev understand canada? here’s a demo where jev predicts the areas of canada that match with a given word it responds instantly and cost me less than a penny

does jev understand canada? here’s a demo where jev predicts the areas of canada that match with a given word it responds instantly and cost me less than a penny

AA

47,445 次观看 • 1 天前

I turned Jev into an image model…? Presenting Jevinci! The paintings are pure JS, no AI image-gen. Jev predicts every pixel's color in parallel and we paint them. The more confident Jev is in a prediction, the wider the brush. (sped up 4x)

I turned Jev into an image model…? Presenting Jevinci! The paintings are pure JS, no AI image-gen. Jev predicts every pixel's color in parallel and we paint them. The more confident Jev is in a prediction, the wider the brush. (sped up 4x)

Anshu

74,182 次观看 • 2 天前

I built a chat bot with jev, no LLM at all! Responses are instant, no hallucinations. I hooked it up to web search, wikipedia, weather, todoist and home assistant. Jev decides what tool to call and what args to use based on the prompt. Instant answers cite sources as well!

I built a chat bot with jev, no LLM at all! Responses are instant, no hallucinations. I hooked it up to web search, wikipedia, weather, todoist and home assistant. Jev decides what tool to call and what args to use based on the prompt. Instant answers cite sources as well!

CJ (Coding Garden)

110,921 次观看 • 3 天前

Okay so Jev can actually do computer use really well Without any screenshots, or LLMs and no Pixels leave my mac I dont even read the Dom elements A local CoreML model segments every button and UI element on screen. On-device OCR reads the labels. That text is all Jev gets. It returns a probability across those elements and tells me the best one to click. Then it clicks, re-runs detection, and decides again. In a loop until the goal is done. ~90ms per decision. Faster than any LLM computer use I've tried. Blazing fast computer use, without any latency TypeSafe AI is building something really interesting

Okay so Jev can actually do computer use really well Without any screenshots, or LLMs and no Pixels leave my mac I dont even read the Dom elements A local CoreML model segments every button and UI element on screen. On-device OCR reads the labels. That text is all Jev gets. It returns a probability across those elements and tells me the best one to click. Then it clicks, re-runs detection, and decides again. In a loop until the goal is done. ~90ms per decision. Faster than any LLM computer use I've tried. Blazing fast computer use, without any latency TypeSafe AI is building something really interesting

Milind S

226,207 次观看 • 3 天前

I got access to Jev. It's incredible. The horizon of what's possible just moved. LLM-era software had a real gap and project liks this was simply impossible. Introducing the first real-time AI-first calculator - no CPU load - instant responses - costs pennies - jev based

I got access to Jev. It's incredible. The horizon of what's possible just moved. LLM-era software had a real gap and project liks this was simply impossible. Introducing the first real-time AI-first calculator - no CPU load - instant responses - costs pennies - jev based

Dmitriy Kovalenko

93,660 次观看 • 1 天前

jev has changed computer-use forever. i built a computer use tool with jev that costs basically nothing to use, and has sub-second latency. it can even decide what to type, and doesn't need an LLM. open source. link below. no LLM required.

jev has changed computer-use forever. i built a computer use tool with jev that costs basically nothing to use, and has sub-second latency. it can even decide what to type, and doesn't need an LLM. open source. link below. no LLM required.

shiv

138,333 次观看 • 1 天前

Played around with TypeSafe AI Jev today, mostly to understand what it does for tool calling. Instead of an LLM deciding what to do, I substituted that part with Jev. My learning is that we will be able to make the agent act before the user finishes the sentence. So far, we've used different LLM combinations (non-thinking + thinking ) plus state machine setups to get the right experience for the end user. TLDR is to understand more about how voice agents can run decisions on partials instead of waiting for the end of the turn, with lower costs.

Played around with TypeSafe AI Jev today, mostly to understand what it does for tool calling. Instead of an LLM deciding what to do, I substituted that part with Jev. My learning is that we will be able to make the agent act before the user finishes the sentence. So far, we've used different LLM combinations (non-thinking + thinking ) plus state machine setups to get the right experience for the end user. TLDR is to understand more about how voice agents can run decisions on partials instead of waiting for the end of the turn, with lower costs.

Pratim Bhosale

110,761 次观看 • 2 天前

Jev is going to be a RL beast for games It took me a week worth of work to get a PPO policy trained to clear Super Mario Bros World 1-1. It took me less than 30 minutes to achieve the same result with TypeSafe AI Jev!

Jev is going to be a RL beast for games It took me a week worth of work to get a PPO policy trained to clear Super Mario Bros World 1-1. It took me less than 30 minutes to achieve the same result with TypeSafe AI Jev!

Shantanu Goel

25,818 次观看 • 3 天前

I've wanted this for ages and it's finally possible now with Jev by TypeSafe AI Imagine Mobbin, but it grabs fresh flows for whatever you’re looking for Browser use has always been to slow to pull this off properly. Jev + browser use are insanely quick now 🫪 I had a lot of fun tinkering on this yesterday

I've wanted this for ages and it's finally possible now with Jev by TypeSafe AI Imagine Mobbin, but it grabs fresh flows for whatever you’re looking for Browser use has always been to slow to pull this off properly. Jev + browser use are insanely quick now 🫪 I had a lot of fun tinkering on this yesterday

omar ✌️

47,574 次观看 • 2 天前

jev is insane 🤯 Here is Jev playing subway surfers at super human speed, and also playing 50 games at once. cost less than a cent to do this run. Jev does not replace llms like astra or fable, but opens up an entirely new world of capabilities.

jev is insane 🤯 Here is Jev playing subway surfers at super human speed, and also playing 50 games at once. cost less than a cent to do this run. Jev does not replace llms like astra or fable, but opens up an entirely new world of capabilities.

Max Blade

345,182 次观看 • 3 天前

I got early access to TypeSafe AI's new Jev model and built a demo of it playing tetris The generation speed is so fast that it's pushing blocks down This feels like another "this changes everything" moment

I got early access to TypeSafe AI's new Jev model and built a demo of it playing tetris The generation speed is so fast that it's pushing blocks down This feels like another "this changes everything" moment

Marcus Lowe

77,960 次观看 • 4 天前

sharing an easy way to test DJev (Diffusion Gemma Jev), the next evolution of the incredible jev concept and fully open sourced (model, code, api...) imagine jev but it also has: - native image-input support - native image as options!! - real-time video streaming input - as fast as jev (not with images, 600ms) - same price as jev, but can get even cheaper - using Diffusion Gemma adapted - we are seeing on-par performance but happy to have external validation! It is a research preview made for fun, so don't expect stability and the awesome scaling as jev team has delivered! this was a weekend project :) have fun!

sharing an easy way to test DJev (Diffusion Gemma Jev), the next evolution of the incredible jev concept and fully open sourced (model, code, api...) imagine jev but it also has: - native image-input support - native image as options!! - real-time video streaming input - as fast as jev (not with images, 600ms) - same price as jev, but can get even cheaper - using Diffusion Gemma adapted - we are seeing on-par performance but happy to have external validation! It is a research preview made for fun, so don't expect stability and the awesome scaling as jev team has delivered! this was a weekend project :) have fun!

David Villalón

18,828 次观看 • 1 天前