Dion

AI specialist | marketer | builder

[Dion @dion0x]

@PokeBurnSol another one

[Dion @dion0x]

@PokeBurnSol No way i won something worthy
How i can get it?

[Dion @dion0x]

@el_hovito yw brotha

[Dion @dion0x]

https://t.co/pmEID6yNNR

[Dion @dion0x]

@WhiteD615590

[Dion @dion0x]

Anthropic just dropped Claude Opus 4.8.

It beats GPT-5.5 and Gemini at coding. But that’s not the big deal.

The big deal: it finally stopped lying. 4x less likely to say “all done” when there are bugs. It actually tells you “not sure about this part, double-check it.”

Numbers?

750,000 lines of code. 11 days. One developer.
Claude Opus 4.8 rewrote the entire Bun runtime from Zig to Rust - hundreds of AI agents running in parallel, 99.8% of tests passing.
Work that used to take a team months.

everyone’s staring at Opus 4.8 benchmarks but the real upgrade is this:

> the model learned to admit when it’s not sure.
> 4x less likely to swear “it works” when it doesn’t.
> an honest model better than a smart model. trust me on this one

[Claude @claudeai]

Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors.

Available today at the same price.

[Dion @dion0x]

@Hazam_502 Fr

[Dion @dion0x]

@2291Sakuya

[Dion @dion0x]

@while_work1 It’s motivating AI race, so that’s good imo

[Dion @dion0x]

@va_tibett Agreed

[Dion @dion0x]

Anthropic just dropped Claude Opus 4.8.

It beats GPT-5.5 and Gemini at coding. But that’s not the big deal.

The big deal: it finally stopped lying. 4x less likely to say “all done” when there are bugs. It actually tells you “not sure about this part, double-check it.”

Numbers?

750,000 lines of code. 11 days. One developer.
Claude Opus 4.8 rewrote the entire Bun runtime from Zig to Rust - hundreds of AI agents running in parallel, 99.8% of tests passing.
Work that used to take a team months.

everyone’s staring at Opus 4.8 benchmarks but the real upgrade is this:

> the model learned to admit when it’s not sure.
> 4x less likely to swear “it works” when it doesn’t.
> an honest model better than a smart model. trust me on this one

[Dion @dion0x]

@simonecanciello GREAT

[Dion @dion0x]

@tar_uniqueee I build similar man. But with stamps

[Dion @dion0x]

@Sayyidalijufri @trust_mrr @dodopayments is ur product better than https://t.co/iR5XhBMmgp?

[Dion @dion0x]

@rossark0 Would be great

[Dion @dion0x]

@WhiteD615590 *yet

[Dion @dion0x]

@WhiteD615590 Just ask claude or another AI, there’s not much info from ppl about it

[Dion @dion0x]

https://t.co/pmEID6yNNR

[Dion @dion0x]

@Amank1412 Guys, I’ve did a research on this topic. Here’s the article.
I hope it’ll help to understand better:

[Dion @dion0x]

https://t.co/pmEID6yNNR

[Dion @dion0x]

@lilyraynyc @Amank1412 Check my recent post, made an article on this topic

[Dion @dion0x]

@JackColbySEO I’ve did a research on this topic

https://t.co/iCY60AMZU4

[Dion @dion0x]

https://t.co/pmEID6yNNR