Anthropic launches Claude Opus 5.5, claiming Fable 5.1-level performance at 40% lower cost than Opus 5.
OpenAI unveils GPT-6 Sol and Luna as faster, cheaper models carrying Astra-era advances in coding and computer use.
Tesla says in-car Grok Connectors can manage inboxes, calendars, and files hands-free.
Vercel publishes fresh Next.js evals: Opus 5.5, GPT-6 Sol, and Fable 5.1 hit 97%; Grok 4.7 reaches 94% at lower cost.
Epoch reports AI inference cost at a fixed performance level has fallen roughly 47% per quarter since 2023.
An Anthropic engineer reports using Opus 5.5 with Lean to generate 16 PRs fixing Claude Agent SDK bugs and races.
Anthropic engineer releases a Claude Code prompt-audit command that finds and removes agent-prompt anti-patterns.
A $25.66, 86-minute one-shot Opus 5.5 run produced an interactive camera-focus and lens simulation.
Theo publishes an Artificial Analysis-based interactive visualizer for agent-model cost/performance comparisons.
Report spotlights allegations that Dartmouth’s provost submitted extensively AI-written work across journals and public writing.