giovanni

gio 🇮🇹 || rs @cognition; prev. @thinkymachines, @GoogleDeepMind, @harvard

"Work hard, study well, and eat and sleep plenty… that is the Turtle Hermit way”

[giovanni @giovanni_]

There is so much value in working dumb that people keep underestimating. Just outwork everything, the answer will become clear with time.

https://t.co/oiPUqBFu8Y

[giovanni @giovanni_]

More intelligence for everyone!!!

Why have to choose? One subscription to try it all with Devin :)

[nader dabit @dabit3]

Cognition is giving away 50 $200 Devin Max plans to celebrate the new model launches! ⚡

Now available in Devin:

• GPT-6 Astra, Sol, Luna + others
• Claude Opus 5.5, Fable 5.1 + others
• SWE-2 (Free until October 15)
• Fusion Frontier harness (Fable, Astra, Sol, Opus)
• Gemini 3.8 Flash + others
• Grok 4.7 + others
• Kimi K3 + others
• Inkling
• DeepSeek V4.1 Flash + others
• GLM-5.3 Flash + others
• Cloud agents on Linux, macOS, and Windows

To be eligible, reply below with what you're building (or something you'd like to build with Devin)! We will choose winners in 24 hours.

[J A Z I I @notjazii]

Grok 4.7 vs SWE 2

tested both models with same prompt at highest reasoning available and results came out really different

> grok 4.7 took 32 minutes and cost $8.14 to make this
> swe 2 took 15 minutes and cost $0 to make this

really surprised by how good a free model is and no idea what grok was doing here

which one did better?

[giovanni @giovanni_]

In the past few weeks people keep texting me pictures of Devin around the world! You now know who is behind it ;)

[Brandon Camhi @bcamhi]

I’ve joined @cognition to lead marketing!

Looking forward to sharing what we’re cooking up. We’re hiring across the board and we’re going to build a legendary marketing team…my DMs are open :)

[giovanni @giovanni_]

@alexsafayan Amen 🙏

[giovanni @giovanni_]

I really believed the world was run by hyper-rationalists, calculating every move in expected value and future power terms. Puppeteers in the shadows who chose the fate of the world.

After a few years, I've realized: no matter how high you go, we're all monkeys.

Same dramas. Same lovers. Same egos.

Humans are really simple creatures and understanding human nature is so important to everything we want to do.

[giovanni @giovanni_]

Congrats, amazing speed of iteration from the @SpaceXAI team!

The story behind Cursor / xAI serves as a cautionary tale for the Valley. Turns out it’s pretty f* hard to reach the frontier, even with all the resources (data, compute, talent…) .

[SpaceXAI @SpaceXAI]

Grok 4.7 is here.

It's a notable improvement over Grok 4.6 at the same price and speed.

[giovanni @giovanni_]

@MollySOShea @Fourthwall @sourceryy Commenting

[giovanni @giovanni_]

@_dan_schwartz AHAHAHHAHAHHAHA

[giovanni @giovanni_]

@brandon_galang @DevinAI Thank you! Feel free to DM any questions / negative feedback as well :)

[giovanni @giovanni_]

@annarmitchell @igarciacamargo Spreading the alpha

[Silas Alberti @silasalberti]

Demand for SWE-2 in Desktop & CLI is unprecedented. Glad to announce, we were able to secure additional compute.

We're making SWE-2 free in Devin Cloud for Pro, Max & Teams subscribers - until October 8. Enjoy!

[giovanni @giovanni_]

@Anqinator 🥹🥹

[giovanni @giovanni_]

“OpenAI awarded us $6,500”

[s1r1us @S1r1u5_]

We reported the bug to Discourse and OpenAI. OpenAI fixed the SSO issue roughly 14 hours after our initial submission.

Discourse received our separate report Saturday, replied Sunday, and had a fix Monday.

OpenAI awarded us $6,500.

[giovanni @giovanni_]

👀🤌

[Anjney Midha @AnjneyMidha]

in the early days of @AnthropicAI

so much had to be done from scratch

it’s a miracle the company exists at all

however, now, so much is off the shelf

entirely new rates of frontier progress

are now within reach for new labs

all you need is the right mindset

[giovanni @giovanni_]

Happy to see our work with SimpleQA pass scrutiny.

Important effort from Epoch: RSI needs good evals!

[Epoch AI @EpochAIResearch]

Introducing Benchmark Reviews: our new initiative to audit AI benchmarks. We are launching with 15 benchmarks: 4 Verified, 9 Flawed, and 2 with not enough information for a review.

[giovanni @giovanni_]

"please fix, make no mistakes"

With Devin + Agentic MapReduce you can finally deslop your code.

[Cognition @cognition]

Some engineering tasks require reasoning across the codebase: Which code is safe to delete? What queries are slowing performance?

Introducing Code Scans: codebase-wide audits for any goal. Devin investigates, reports findings, and opens the PRs. Powered by Agentic MapReduce.

[Russell Kaplan @russelljkaplan]

Mac VM support is so fun. Uploaded some baby pictures and converted them to video game sprites for Devin to make Flappy Asher. Live in TestFlight. My wife is thrilled.