> But with no benchmarks or price comparison, which likely means it doesn't compare that well.

Doubt that's the case; this was only the announcement and the API isn't even released yet. As I noted, my own testing using a batching strategy (nothing more than "read 20 lines from this file instead of reading 1 at a time") yielded ~the same accuracy on an 800 row dataset we have with ~1.6x the throughput (at only batch_size=20) and 1.2x the cost. Didn't tune to find accuracy dropoff by batch size, but it's easy to see that the decision capability was there (understood that it's not the same without the probabilities, but even without it, the classifications were nearly identical).

Investors/founders are likely horrified to discover there is no industrial-scale moat in no time at all... they probably expected to have at least a 2-3 month window to capture market share and value.

I suspect Jev was intended to be an acquisition play and instead OAI said "nah, we're good" literally 2 weeks to the day after its launch. Anthropic and the Chinese labs are almost certainly soon to follow.

They're rumored to be in the early stages of a billion dollar round to value the company at $10B. Best of luck with that.

Validated, envigorated ?

The reaction to Jev has been huge, as has been OpenAI's speed here is announcing a "not quite ready yet" competitor... the idea of Jev is one of those "Doh! Why didn't I think of it!" moments ... in retrospect it's obvious that business automation is all about decision making and the market for this is massive. Up until now most people just saw this as a use case for LLMs, until TypeSafe/Jev came along and said "actually you don't need an LLM for this ...".

I'm sure TypeSafe realized the idea is a winner and expected competition, although maybe not quite so fast?

The question now is how seriously OpenAI take this, and are they willing to compete with Jev on price, thereby massively reducing Luna margins (which OpenAI's "Decisions API" is based on)? I guess if you can't make $0.10/M for Luna, then $0.04/M is better than just giving that market to a competitor, assuming you're profitable at $0.04/M.

Flattered, I assume.

How is it not a rugpull?

They halved the usage of existing Pro 20x subscribers, then created a new plan with the old usage.

Just because it's not literally illegal doesn't mean it's not a rugpull.