Someone gave Canva access to GPT-6 and Fable 5.1.
One destroyed the other (and it wasn't even close). Claude Fable 5.1 is on the left. GPT-6 Astra is on the right. The difference is not small. It all comes down to one test: ARC-AGI 3. This test places a model within
This started with one simple prompt
Claude Opus 5 took the idea and ran with it. A few iterations later… This was the result.
Fable 5.1 cost about 40% less than Opus 5 across these six tasks, despite higher input and output rates.
Compare the whole job. Full comparison on YouTube: Amir Zaji. All my free resources:
Inception launches Mercury 2.5, the smartest diffusion LLM yet
Today Inception announced Mercury 2.5, a diffusion language model the company calls its most capable yet. According to Inception, it delivers a 40% intelligence jump over Mercury 2 while still running at more than
You have reached the end of the archive
All of Claude Opus 5