Rami Alhamad

I like to build things.

[Rami Alhamad @RamiMatic]

@cohere Let’s go 🥳

[Rami Alhamad @RamiMatic]

Custom Alma 2.0 icon vs Spotify Disco ball?

[Rami Alhamad @RamiMatic]

Alma 2.0 is out and thought to take a different take on Dark Mode, introducing Kale Mode (TM) 🥬

[Rami Alhamad @RamiMatic]

@DanielMiessler “Help rephrase this policy change in a confusing and obscure way”

[Rami Alhamad @RamiMatic]

@dedene @AnthropicAI The gaslighting is somehow worse than the policy change

[Rami Alhamad @RamiMatic]

@clairevo @danshipper 😂 nailed it

[Rami Alhamad @RamiMatic]

@ClaudeDevs Claiming this is a perk is insulting to your Max users. I'm seriously considering unsubscribing.

[Rami Alhamad @RamiMatic]

@alexgreensh @ClaudeDevs @lydiahallie 100% agreed

[Rami Alhamad @RamiMatic]

@theo @ClaudeDevs Ditto, this is insane that @AnthropicAI is trying to spin this as a win for developers.

[Rami Alhamad @RamiMatic]

Seeing what AI is doing to content these days really reminds of this infamous clip: Everything is a Remix.
https://t.co/5We57APKGR

[Rami Alhamad @RamiMatic]

@higgsfield has the most aggressive upsell strategy I've ever witnessed.

[Rami Alhamad @RamiMatic]

@robjama 😂 I literally have Codex Plus + CC Max side by side at this point. I'm finding Codex though is starting to tip over in usage.

[Rami Alhamad @RamiMatic]

@alex_whedon Signed up! Looks really interesting

[Rami Alhamad @RamiMatic]

🤯

[Alexander Whedon @alex_whedon]

Introducing SubQ - a major breakthrough in LLM intelligence.

It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA),

And the first frontier model with a 12 million token context window which is:

- 52x faster than FlashAttention at 1MM tokens
- Less than 5% the cost of Opus

Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention).

Only a small fraction actually matter.

@subquadratic finds and focuses only on the ones that do.

That's nearly 1,000x less compute and a new way for LLMs to scale.

[Rami Alhamad @RamiMatic]

What effort levels actually mean in @claudeai

[Karri Saarinen @karrisaarinen]

The obvious path for designers in the AI age is to move closer to code.

But the more valuable path may be upstream: closer to the customer, the business, and the problem.

If everyone can prompt agents to code, the scarce skill becomes knowing why, what, and how to build.

[Rami Alhamad @RamiMatic]

Credit where credit is due, @OpenAI Codex has come a long way. I'm now on their Pro tier too.

[Rami Alhamad @RamiMatic]

@grok 4.1 Fast is 3x slower than an old GPT 4.1 model?

[Rami Alhamad @RamiMatic]

@GergelyOrosz Coders, yes, builders, no.

[Rami Alhamad @RamiMatic]

Ok GPT 5.5 is pretty impressive, cracked a UI issue that Opus 4.7 kept going on circles on.