Akshit

looking at data™️ @ProximalHQ
interested in evals, interpretability, monitoring

[Akshit @akshitkr]

@kreepkroop how did i miss this that is crazy

[Here's What I Reckon: @angryaboutbikes]

The cool thing is that you actually can, for a while, until suddenly a very minor difficulty causes you to have a breakdown and you lose the ability to function as a human on any level for months or years

[Joe Standoff @wetworkarchive]

They don't work because the absolute core of ADHD is emotional dysregulation, which is another way of saying a diminished sense of internal self. All humans learn internal regulation first by it being modelled to them, and many people with ADHD never had the model.

[Akshit @akshitkr]

@Ninelxarm @LagrangianSad mb bro

[henry @arithmoquine]

remember this old post of mine? Tried it with Jev---it seems to have a grasp of the globe on par with some of the best models a year ago

due to the architecture & low cost, it was also feasible to extract a labeled map of continents and countries. lots of interesting details

[Akshit @akshitkr]

@Adedxpo @OneSodiqAlabii Foucalt talks about this

[Hacker News @hackernews]

A camera that only sees one column of pixels at a time turns out to be the perfect way to photograph a streetcar.

Daniel Lawrence Lu stood by the Muni tracks for three hours with a line scan camera, capturing about 10,000 one-pixel slices a second as each tram rolled past.

Since the vehicle paints itself onto the frame, there's no perspective at all. The trams come out flat and dead-on, like the elevation drawings an engineer would make, while the background smears into stripes. And people don't blur either.

https://t.co/Txt5Ij05hd

[Akshit @akshitkr]

@railmaxxer this stretch is a car centric mess and gets clogged up every evening

[Akshit @akshitkr]

i just realised astra is 100x the cost. im sure there are thousands of workflows using Astra which could do just fine with Luna, all downstream of having good internal evals

[OpenAI @OpenAI]

Please welcome GPT-6 Sol and GPT-6 Luna to the GPT-6 universe.

GPT-6 Sol and Luna build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale.

We’ve also made caching and inference more efficient, and we’re passing the savings directly to you: 50% lower API prices for Sol and Luna compared with GPT‑5.6 promotional pricing.

[Akshit @akshitkr]

@ShriramKMurthi curious what had to change for you to come this side of the fence?

[Leon Lin @LexnLin]

every frame of this is painted with JavaScript

made with Opus 5 (probably routed to 5.5),
sounds on :)

https://t.co/vmz4CxmIff

[Proximal @ProximalHQ]

Opus 5.5 is both the cheapest and highest-scoring model released by Anthropic

This trend has been constant since Fable 5 - every model released after it scored higher while being cheaper than its predecessor

[Proximal @ProximalHQ]

We partnered with @FireworksAI_HQ to bring FrontierSWE v2 to the Specialized Intelligence Index

Evals are essential for improving and safely deploying AI. We're excited to support the initiative!

[Chris Barber @chrisbarber]

I asked people: "which data companies should an ambitious future founder join?"

13 votes: Mercor (@BrendanFoody @adarsh_exe @suryamidha)

11 votes: Fleet (@nicoup @andrewthezhou)

10 votes: Proximal (@MatternJustus @navidkpr @calvinchen)

9 votes: Surge (@echen)

8 votes: AfterQuery (@CarlosGeorgescu @spencermateega)

7 votes: Mechanize (@GuiveAssadi @tamaybes @EgeErdil2 @MatthewJBar)

6 votes: micro1 (@aliansarinik)

4 votes: Datacurve (@serenaa_ge @charleyslee)
4 votes: OpenAI (data team) (@BarrAlexandra @phoebethacker)
4 votes: Preference Model (@Ning_Catsnail)

3 votes: Bespoke Labs (@madiator @AlexGDimakis)

2 votes: Anthropic (data team)
2 votes: Datology (@arimorcos @hurrycane @leavittron)
2 votes: Long Lake (@alextaubman @rasmuswissmann @varunshenoy_ @tarasholovko)
2 votes: Sieve (@mokshith @thpicy)

People couldn't pick a company they work at or founded. Disclosure: I'm a small investor in AfterQuery and Fleet (in this post), and in Abundant, Applied Compute, Latch, Olam Labs, Trajectory and Vmax (elsewhere on the list). I didn't vote.

[Akshit @akshitkr]

@ananayarora Congratulations!

[Epic Mike @The_Epic_Mike]

The first season of True Detective might actually be the greatest season of television ever made.

[tommaso @sillynous]

oh my god

[Akshit @akshitkr]

@LagrangianSad we got caste 2

[Justus Mattern @MatternJustus]

We are hiring a generalist intern this fall to work closely with me on applied research!

You should be technical, but you will not be asked to write code all day - instead, we will work together across product, operations and ensuring our customers are happy

DMs open!