Igor Manturov Jr.

Erlang and Swift

[Igor Manturov Jr. @AchillesPicture]

@GenMagnetic It's awesome!

[Igor Manturov Jr. @AchillesPicture]

@emanperez28 Interesting, but I feel like my arm would get tired way faster than just using a trackpad. Seems pretty inconvenient to me.

[Igor Manturov Jr. @AchillesPicture]

@elefantopotamus Codex internal browser

[Igor Manturov Jr. @AchillesPicture]

AGI has arrived.
Maybe Astra is great at generating useless 3D websites, but it successfully failed at the most basic real-world task: registering me on a visa portal.

I’m not even talking about filling out the visa application. It spent 5 minutes failing to simply enter an email address for registration.

So much for AGI..

On an actual practical use case that should provide real value to the end user complete failure.

[Igor Manturov Jr. @AchillesPicture]

@AlanTheKnight @thsottiaux @ChatGPT @OpenAI Chile 199 990 CLP ($216 USD)

[Igor Manturov Jr. @AchillesPicture]

@elefantopotamus https://t.co/DwfFVmwktF

[Igor Manturov Jr. @AchillesPicture]

@christitustech same! 4% of weekly usage per hour with Astra Medium on a $200 plan. (96% 24 hours)

[Igor Manturov Jr. @AchillesPicture]

@buildwithrajath 4% of weekly usage per hour with Astra Medium on a $200 plan. That’s 96% of the weekly allowance in just 24 hours.

[Igor Manturov Jr. @AchillesPicture]

@derrickcchoi 4% of weekly usage per hour with Astra Medium on a $200 plan. That’s 96% of the weekly allowance in just 24 hours. Coding tasks!

[Igor Manturov Jr. @AchillesPicture]

@hqmank I didn't notice. 4% of weekly usage per hour with Astra Medium on a $200 plan. That’s 96% of the weekly allowance in just 24 hours.

[Igor Manturov Jr. @AchillesPicture]

@DanDr1s my flow: 4% of weekly usage per hour with Astra Medium on a $200 plan. That’s 96% of the weekly allowance in just 24 hours.

[Igor Manturov Jr. @AchillesPicture]

@christitustech 4% of weekly usage per hour with Astra Medium on a $200 plan. That’s 96% of the weekly allowance in just 24 hours.

[Igor Manturov Jr. @AchillesPicture]

@melvindvivas 4% of weekly usage per hour with Astra Medium on a $200 plan. That’s 96% of the weekly allowance in just 24 hours.

[Gary Marcus @GaryMarcus]

“The expected value of a fully LLM-generated, non-human-reviewed patch is a net-negative by a considerable margin.”

[Igor Manturov Jr. @AchillesPicture]

@sethrose @thsottiaux Sol Medium as the orchestrator + Sol Medium as the executor + Sol XHigh as the reviewer burns about 1% of my weekly ProX20 usage per hour of continuous work. The exact same setup with Astra burns about 4% per hour.

[Igor Manturov Jr. @AchillesPicture]

@pvncher Please fix Astra’s token consumption. Even on Low and Medium, it’s practically impossible to use on a $200/month subscription.

[Igor Manturov Jr. @AchillesPicture]

@karlmehta OMG! It's AGI!

[Igor Manturov Jr. @AchillesPicture]

@thsottiaux @TJeparskis Sol Medium as the orchestrator + Sol Medium as the executor + Sol XHigh as the reviewer burns about 1% of my weekly ProX20 usage per hour of continuous work. The exact same setup with Astra burns about 4% per hour.

[Igor Manturov Jr. @AchillesPicture]

@hany99dev Sol Medium as the orchestrator + Sol Medium as the executor + Sol XHigh as the reviewer burns about 1% of my weekly ProX20 usage per hour of continuous work. The exact same setup with Astra burns about 4% per hour.

[Udit Goenka @iuditg]

OpenAI should stop hyping their model.

Calling GPT 6 Astra as something close to AGI is wrong.

It still sucks to manage coding tasks.

- Continues to over engineer.
- Continues to write poor quality code, and requires babysitting.
- Continues over spending tokens on useless loops.

As of now OpenAI’s model doesn’t come close to Fable series.

The difference is massive.

As many of you might know from my previous posts, I am not a huge fan of Anthropic as a company, but if we purely talk about technicality, and getting a task done…

Fable is cheaper, gets shit done, and understands coding tasks so much better.

I do not care about 3D modelling etc. so if those are the areas Astra has worked on then it’s a useless update for me personally.

As of now, the old issues that were reported in GPT 5.6 Sol series continues.