Stephen M. Walker II
HomeBlogLinksAdventuresAboutContact

Google continues to dominate benchmarks with its frontier Gemini 2.5 Pro 0605 model. Smarter, faster, and cheaper than Anthropic models.

Just missing a Gemini Code package.

Compared to OpenAI's GPT-4.1, Gemini 2.5 Pro is a bit too eager and ignores specific edit instructions, despite acknowledging in the CoT.

Published on June 5, 2025