Alex Atallah
@alexatallah
CEO, co-founder @OpenRouter, the largest LLM marketplace. Previously: co-founder & CTO @OpenSea.
3.4K Following 67.1K Followers
The Jev Router is our most transparent router to date, and it benchmarks very well!
See how it picks the model in the Chatroom sidebar, including cache-aware cost tradeoffs:
Very good chance that decision models unlock the next breakout AI consumer app, something that's been notably missing since Character AI.
AI will power some new games that were never feasible before!
.@openrouter tip: you can get useful notifications like these directly in your team's Slack
Curious what Jev's favorite ____ is?
Try to get your thing to rank first!
You can also play with Jev directly in our playground:
Quite an interesting turn in LLM diversity, if the market starts looking for specialized models that are optimized and branded for different use cases.
Should be very good for the market and open more room for non-frontier neolabs!
Show more
new york jeveloper night just got bigger: @OpenRouter is providing credits for attendees!
we’re capping the room at 150, and @jjacky will share how he’s been working directly with the @typesafeai team.
6pm this thursday at @betaworks.
demos + food + drinks.
register:
Show more
The argument over Anthropic in the FT today
a) Under pressure from OAI, open source, all fronts
b) the lynchpin of the AI bull case
Featuring @OpenRouter data of course :)
Show more
What is a decision model?
Jev by @typesafeai 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:
Show more
Jev by @typesafeai is the first model ever to truly optimize for low- and medium-entropy tasks
I tell many of our customers that they should be decomposing their AI tasks into three buckets:
1. Low entropy: e.g. "who should review this PR?"
2. Medium entropy: e.g. "review this PR"
3. High entropy: e.g. "write a PR"
Today, I think you only need frontier models for #3#!
Show more
Jev is a model that efficiently makes */decisions*
This is a huge devex win for AI. As a team that believes in ruthless type checking, being able to expand this love to our classifiers and decision making processes means 100x more AI usage.
The quintessential Jevons paradox model
Show more
Jev by @typesafeai is now on OpenRouter, in beta.
Jev is a System One model. Instead of generating text, it takes your app's state plus a typed question and returns a typed decision with a probability attached. There is no JSON prompting, parsing layer, and nothing to validate against.
Show more
cool feature update from @OpenRouter, I think it's an underrated step in providing a one stop shop to building your own production agent.
Right now it allows bash commands and file uploading, but I think an opportunity is server side REPL/code mode programmable tool calls.
Any tool call you define on your app should be passable to OR server side so that it can be executed in the script, compute and reduce context to the model, reducing costs, increasing capabilities, and improving latency.
Show more
1/ Any model on OpenRouter can now run code in a hosted Linux container.
Add one tool to a Responses API request. The model writes the script, runs it, and reads back the result. Nothing to install, nothing to host.
Meet openrouter:shell.
To get started:
"tools": [{ "type": "openrouter:shell", "parameters": { "engine": "openrouter" } }]
Show more
Here's the DeepSWE result for Union Alpha, a stealth model we just launched.
Try it now! Works in every harness.
$ ori [code | your-fav-harness] --model=stealth/union-alpha
I tell many of our customers that they should be decomposing their AI tasks into three buckets:
1. Low entropy: e.g. "who should review this PR?"
2. Medium entropy: e.g. "review this PR"
3. High entropy: e.g. "write a PR"
Today, I think you only need frontier models for #3#!
Show more
OpenAI had not taken #1# in wallet share on OpenRouter for 133 weeks - but Astra pushed them into the top spot last week.
OpenRouter users spent more on OpenAI models than on Anthropic models last week.
This hasn't happened for more than 2.5 years
OpenRouter users spent more on OpenAI models than on Anthropic models last week.
This hasn't happened for more than 2.5 years
Got married last week and an @OpenRouter key was definitely the last thing I expected to find in a wedding card.
Cool gift though!
OpenRouter has the most rigorous integration process of any router
I love that they’re so focused on making sure they have the very best quality on the platform and they push providers to get better (including us!)
Great for users, great for the market, long live OpenRouter
Show more
I tell many of our customers that they should be decomposing their AI tasks into three buckets:
1. Low entropy: e.g. "who should review this PR?"
2. Medium entropy: e.g. "review this PR"
3. High entropy: e.g. "write a PR"
Today, I think you only need frontier models for #3#!
Show more
