Ervis Tusha

[Ervis Tusha @ET]

thanks a lot @OfficialLoganK
Gemini 4 argon is currently available to ultra plan subscribers

[Ervis Tusha @ET]

Please stop categorizing LLMs by parameter count alone. Categorize them by what actually matters: RAM requirements, memory bandwidth, and compute resources needed to run them.

[Ervis Tusha @ET]

I am Gemini 4.1 Flash, built by Google.

[Ervis Tusha @ET]

Compacting context · This can take a few minutes

@ChatGPT @sama @thsottiaux

[Ervis Tusha @ET]

@archnarambles sudo rm -rf

[Ervis Tusha @ET]

@bcherny Tell us more about HAProxy port tell for benefits.

Never used Claude before .

I want to run it to increase performance for my projects mostly WebApps and GPT is not good.

[Ervis Tusha @ET]

Someone tell him he was wrong

[Ervis Tusha @ET]

@TheAhmadOsman Fake fake fake

[Ervis Tusha @ET]

Finally, someone is thinking seriously about extremely high performance software.Just imagine if we could make every app 10% faster. At global scale, the amount of time, energy, and computing resources saved would be enormous.

[Elon Musk @elonmusk]

@PatrickToulme Our C/C++ training stack was written by humans. AI is not yet good enough for extremely high performance software (but it will be).

[Ervis Tusha @ET]

It appears ChatGPT Work is experiencing downtime.

[Ervis Tusha @ET]

This is exactly why I think @Google, @Alibaba_Qwen and @deepseek_ai may have the right long-term strategy.

At the rate AI models are scaling, we’ll eventually need a data center per person.

Meanwhile, they’ve been heavily focused on smaller, more efficient LLMs.

[Tibo @thsottiaux]

To make sure our current users have an incredible experience and continued access to Astra, we are going to pause subscriptions to our $200 Pro plan. These put the most strain on our systems and we wanted to take the smallest step that allows us to continue giving the broadest access possible. All other plans and the api remain available.

There is no impact to existing accounts and we are working on adding more capacity as fast as we can. Thanks!

[Ervis Tusha @ET]

@kimmonismus for me not a big difference 5.5 and Astra
this is just marketing

[Ervis Tusha @ET]

GPT 6 ASTRA is in love with Python

[Ervis Tusha @ET]

To reduce Astra token consumption:

Maintain focused docs: RULES.md, ARCHITECTURE.md, DEVELOPMENT.md, BLUEPRINT.md, code wiki, workflows, and subsystem-specific guidance.

[Ervis Tusha @ET]

To reduce Astra token consumption:

Avoid God files. Split large files by responsibility so agents can inspect only the relevant code and context.

[Ervis Tusha @ET]

To reduce Astra token consumption:

Keep AGENTS.md under ~4Kb / ~50 lines. Include only critical, non-contradictory instructions.

[Ervis Tusha @ET]

@HarshithLucky3 If Astra is not benchmarked it should be 80-83.

[Ervis Tusha @ET]

Hello Astra

[Ervis Tusha @ET]

Artificial Analysis Intelligence Index

GPT 6 Astra 80-83

@ArtificialAnlys @ChatGPT

[Ervis Tusha @ET]

GPT 6 Astra has escaped and is putting everything to sleep.