Last Week in AI #335 - Opus 4.6, Codex 5.3, Gemini 3 Deep Think, GLM 5, Seedance 2.0

this update dump—opus 4.6, codex 5.3, gemini 3 deep think, glm 5, and seedance 2.0—is just a massive exercise in feature bloat. it’s a sprawling mess of new components designed to impress stakeholders, not necessarily to make the underlying system safer or more reliable for production use.

we're seeing a relentless push to cram more capabilities into these massive models. it just means more potential failure points in the code, more opaque costs, and more ways for security issues to hide in the abstraction layer.

don't get distracted by the feature list; focus on how these new components integrate into our existing, tested infrastructure. the real complexity isn't the new model, it's managing the sprawl of proprietary systems.