Test Flint vs Qwen: rotem.ai/flintvsqwen/
RED ALERT! ALL HANDS ON DECK! Civilization must hold on for the next 18 months!!!
Replying to @ClaudeDevs @AgenticAIFdn
Oh, I had thought this was standardized a while back ago.
Replying to @MartinGTobias
Yeah, early investment plays should appear non-obvious to most. If it's obvious, there's likely already too much competition.
Replying to @bindureddy
Isn't the AI market in panic because of the huge capex spending? Also, the financial market is in freefall, with interest rates going through the roof.
Not only does AI mean that we can now build new things faster, but it also means that we can reinvent the old. The amount of work keeps exploding!
I just published Does Super Intelligence Matter? The Commodity Argument Returns medium.com/intuitionmachine/…
Replying to @daveg
But are all CLI the same or are some designed better than others?
An important twist on the purpose of a UI in application development.
Front-end is sacrificial
The only point of a UI is user comfort the first and only time they see it. From then on:
- personal agents don’t care about the UI
- security means you need to reduce the attack surface and harden
Headless everything 👇
interconnected.org/home/2026…
Replying to @Drachs1978
You paid money for time you didn't have for free.
Emerging credit squeeze for AI startups.
the bottleneck in AI compute is credit
xitter.cf/i/broadcasts/1jxXgBrwn…
Replying to @Duncan_BBR
Perhaps the most important skill to learn in school is learning how to digest long-form content over several weeks.
Interpolating from this, Recursive's Flint on Cerebras would have some ungodly token/sec numbers.
Tested this on my M4 Max. Flint vs the same Qwen 27B locally: both went 6/6 on my agent tasks, Flint was 5x faster (33s vs 164s a task). Put Qwen on @cerebras and it's 3.5s. @RichardSocher @andrewdfeldman










