Juri Ganitkevitch

NLP reply guy.

[Juri Ganitkevitch @thejuri]

What did you think pacing the frontier meant? Vibes? Essays?

[Tibo @thsottiaux]

Hi,

Tomorrow we are re-opening the Pro $200 subscriptions to new subscribers, but together with it we are also changing how we calculate the usage for it. In effect, if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan.

Now that it's said, let me explain why this is happening and why you will still get more work done than if you were on the Pro $200 subscription one month ago.

(a) We didn't want to compromise in other ways and are committing to not reintroducing the 5h limit, so that you can fully use the weekly usage when you want.
(b) On the subscription, we guarantee that over time you always get more work done and with an increasing level of quality. This means that you will continue to get more value per dollar spent as a result of models getting more efficient and us passing down the improvements in the form of API price reductions.
(c) We don't want to put an incentive on ourselves to artificially inflate the API list prices to make it look like you are getting a lot (and workaround it through discounts, etc). Instead we want to continue to both rapidly reduce prices and increase capabilities of models on the API. This week we introduced GPT-6 Sol and GPT-6 Luna at 50% of their previous price. Over time, we see prices go low enough that it makes sense for most to buy usage as needed without there being a significant gap between what you get in a subscription and what you get in the API for a dollar spent.
(d) Tomorrow, we are adding more things to the subscription that won't draw on the usage, I won't reveal what that is yet.

I wanted to be transparent before all the big announcements tomorrow. Lots of new exciting things are coming to the subscriptions that will make it super compelling, but I wanted to make sure to share this change ahead of time so you can all understand it before we shower you with good news.

Codexingly,
Tibo

[Juri Ganitkevitch @thejuri]

RT @ben_lern: Now you can save Snowflake credits... with Snowflake credits.

Espresso is available via snowflake's marketplace, which means…

[𝐟𝐚𝐮ᛋ𝐭 @faustianwigger]

"It's not our war" well they just hit one of our embassies so I think it's starting to be our fucking war

[Juri Ganitkevitch @thejuri]

“what’s the denominator!?” i scream as the token vol share charts crowd around me, more joining in every moment. they pack me into the back of an unmarked van. “but the selection bias..” the van drives off.

[Nathan Lambert @natolambert]

Here's what the historical open vs closed data here looks like. Would be useful to have volume too (not just %)!

[third-party evaluation partner @mycoliza]

it’s not “hyperefficient” it just sucks man

[Omar Khattab @lateinteraction]

my days are like `N` hours of working with frontier models, babysitting them in frustration and disbelief at the frequent errors and slop

interrupted by `M` min breaks of browsing X takes about how the same models (and _especially_ the next gen!) are superhuman at ~everything

[Edoardo Contente @edoardocontente]

There is a super shady data control setting on everyone's ChatGPT which makes it seem like your data can be used for training even if you explicitly say to *NOT* improve the model for everyone (1/9)

[Juri Ganitkevitch @thejuri]

We're building this at @espresso_ai! Verified hypercompiler optimization, models that understand load scaling & boundedness, plus load-forecasting aware scheduling. Cut our own AWS costs in half last month.

[Jay Kreps @jaykreps]

How much longer before this kind of automated optimization can be done reliably at the AWS account level across a large application footprint? A sad fact of the cloud, as with all infrastructure, is most customers run at about 15% utilization.

[Ben Lerner @ben_lern]

@jaykreps Soon! We do this for data warehouses - it works better than we expected, and extending it to Spark was easier than we expected. I think general-purpose tooling is around the corner.

[Ben Lerner @ben_lern]

@LuxuryWatchGuy1 @sama coincidentally I, too, have ordered a set of ultra-rare custom watches fully branded for our team

[Tomás Bjartur @BjarturTomas]

From the OpenAI HuggingFace hack report:

[Juri Ganitkevitch @thejuri]

@ben_lern This isn't just good advice – it's transformative!

[Mengye Ren @mengyer]

To lower the submission volume and reviewer burden, charge
- $100k fee to submit a paper.
- $1M gold status for unlimited number of paper submissions for a year.
- $5M platinum status for unlimited paper acceptances for a year.

[Espresso AI @espresso_ai]

Exciting news - USA Today covered Espresso AI’s launch of Kubernetes for Snowflake 🚀

https://t.co/VmWDtQ66hK via @usatoday

[Ben Lerner @ben_lern]

Espresso's in USAToday!

https://t.co/CLkrEKCPpM

[ᐱ ᑎ ᑐ ᒋ ᕮ ᒍ @Andr3jH]

thank you mistow pwesident. thank you fow youw leadowship. thank you for being the best pwesident in histowy. thank you. thank you. thank you. thank you. thank you. thank you.

[(((ل()(ل() 'yoav))))👾 @yoavgo]

if i read things correctly, OpenAI realized that you can get by with much fewer layers if you compensate by generating more tokens ("thinking"). this is huge as it offloads a lot of the cost to the consumer: each gen step is now cheaper to produce, and they sell more of them.

[Espresso AI @espresso_ai]

Espresso AI Launches Kubernetes for Snowflake to Renovate Data Warehouses
https://t.co/fyfsq4MeTt

[Markov @MarkovMagnifico]

the xai team is so disingenuous every time they post one of these apologies

we know they manipulate the information stream and steer grok’s behavior for political purposes. it’s heavy handed and it’s obvious — I don’t know why anyone who respects themselves would work here