Jack

CEO/Beamo

[Jack @jackkallisonn]

You're 5 minutes of work away from having Linux on your computer - no experience required.

[Jack @jackkallisonn]

@ImSh4yy Fable 5 launch had the most aura of any launch ever and the US government banning it only added to that. They lost it all cause they have no compute.

[Jack @jackkallisonn]

@MatthewBerman Kimi's also a new pretraining run as well. I honestly think they have almost fully caught up.

[Jack @jackkallisonn]

@tomhacks Yeah now I can just keep my codex sub for everything except frontend and a small kimi sub for frontend so that I don’t have to pay for Claude Code anymore

[Jack @jackkallisonn]

@lyter_r @AlexFinn Yeah and you can't even use subagents, only one at a time 😭

[Jack @jackkallisonn]

@adrian1977 @AlexFinn Yeah exactly like you can run stuff better than original gpt-4 on even your phone now but by today’s standards gpt-4 is really bad, you’re always going to want the frontier cloud stuff

[Jack @jackkallisonn]

@ParsaKhaz I actually made an easy open source version to do this easier and it works with skills on Codex and Claude and everything. It’s so much better than the manual setup. https://t.co/cHOHGIz3wD

[Jack @jackkallisonn]

Just dropped a completely free and open source Codex in Claude Code harness that's easy to use and has complete feature cross compatibility. Skills from Claude and Codex work great with no issues. If you want to use it or view it on GitHub, you can at https://t.co/cHOHGIz3wD

[Jack @jackkallisonn]

@Kimi_Moonshot Can we use it now?

[Jack @jackkallisonn]

@wavefnx It's a bad deal but it's actually pretty useful for managing a bunch of stuff at once

[Jack @jackkallisonn]

@blueemi99 Realistically it's probably GPT 5.5 level but cheaper which is still really impressive and would make it by far and away better than GLM 5.2

[Jack @jackkallisonn]

@kr0der It's actually so weird thinking about how many people there are that use Codex and Claude without X 😭

[Jack @jackkallisonn]

@AmericaAeterna @_Meshak @ThePrimeagen @arena Yeah ik what arena is, it just judges what responses people prefer when it comes to tone and how it sounds to the user, not actual intelligence. Anthropic models are obviously the best to talk to so they win this but Fable and 5.6 are pretty much tied on all of the coding benches

[Jack @jackkallisonn]

@willccbb Clicky ofc

[Jack @jackkallisonn]

@AmericaAeterna @_Meshak @ThePrimeagen Opus 4.7 above GPT 5.6 I'm really sure this is a great benchmark

[Jack @jackkallisonn]

@thsottiaux It's so much better without the 5 hour limit. I'm actually able to continuously work on everything without being stopped or interrupted when I need a lot of usage in a quick burst.

[Jack @jackkallisonn]

@Kimi_Moonshot 333333333

[Jack @jackkallisonn]

@thsottiaux Does the Codex Micro come with a free reset inside the box?

[Jack @jackkallisonn]

@theo 9M sub milestone almost here

[Jack @jackkallisonn]

@jun_song We can’t be serious right now. Deepseek V4 is nowhere near Opue 4.8 lol