Jaid

0.30000000000000004× full-stuck slopware engineet

[Jaid @JaidCodes]

The Great Subsidization arc is officially over.
No more Copilot shilling.
No more flatrate exploit showcase.

It’s time for the tokenminning arc.

[Jaid @JaidCodes]

@Kikobeats snake oil final boss

[Jaid @JaidCodes]

failed state of the art

[Jaid @JaidCodes]

@ivanfioravanti Very confusing name.

It’s meant to assist Gemma (by narrowing the scope of possible next tokens during generation), not the human. It’s also 400 million parameters, not 12 billion.

[Jaid @JaidCodes]

@Jaaneek So is xAI fine with using the Grok Build OAuth client from third-party harnesses in general (in an honest way, so with a custom User-Agent set)?

Or is your tolerance limited to those three clients (OpenCode, OpenClaw, Hermes) specifically?

[Jaid @JaidCodes]

I was digging into this and I’m amazed by the way it works.

OpenCode doesn’t have a valid OAuth app, so they use the sign-in flow for Grok Build. Same as Hermes and OpenClaw.

[Jaid @JaidCodes]

I’m absolutely going to mimic the OpenCode client in order to use my Twitter account’s included Grok quota from my own harness.

[Jaid @JaidCodes]

@Hesamation It still is.

[Jaid @JaidCodes]

Somebody accidentally poked his eye with a fork, so everybody should be eating with bare hands.

[Jaid @JaidCodes]

@SasaMarinkovic @AMD Why does it say “30 billion” for a 35b model?

[Jaid @JaidCodes]

It’s as official as it gets.

All three integrations received direct contributions from an xAI employee (@jaaneek).

[Jaid @JaidCodes]

It’s not an uncommon practice, I just didn’t expect to see it in this particular case because of how official the xAI subscription support seems.

[Jaid @JaidCodes]

@pmarca WTF I love water now

[Jaid @JaidCodes]

@hthieblot

[Jaid @JaidCodes]

@_julianschiavo

[Jaid @JaidCodes]

It’s regarded. 🥀

[Jaid @JaidCodes]

Update on this:

I went looking for some simpler models that are working out of the box and still match my criterias (20–40 billion parameters, dense, agentic, preferably not multimodal, preferably not open-source) and found this one.

Seems perfect for my current purposes.

[Jaid @JaidCodes]

Burnt $ 100 on tokens today, got barely any value out of it, decided to take a small break from cloud inference, asked GPT-5.5 to prepare 8 different inference server setups for my local rig, all of them failed.

[Jaid @JaidCodes]

I made a container for it and it runs well with no further custom preparation step needed.

[Jaid @JaidCodes]

@kentcdodds worth the new pricing

[Jaid @JaidCodes]

@krea_ai @ArtificialAnlys Where woman lying on grass?

[Jaid @JaidCodes]

@pleometric @nikitabier

[Peter Dedene @dedene]

POV: you're still using GitHub Copilot after June 1st, 2026