News
AI agents, model releases, and industry study
• Safely scaling multi‑agent teams with Claude Fable 5.2 – 1,000 agents proved Fermat’s Last Theorem in 28.7 hours • Claude Fable 5.2 and Mythos 5.2 pricing and capabilities – $8 / M input tokens, $40 / M output tokens, 30% faster generation • PostHog decision‑model approach reduces LLM calls by 98.7% • Dropbox AI‑assisted development delivers 22 accessibility fixes and clears 53 test tasks in a week • Anthropic model guide: Opus 5.5 for most work, Fable 5.2 for hardest tasks • Meta’s Muse Ember multimodal model offers strong cost efficiency • Payroc‑commissioned study shows slow payment integrations cost revenue, with 12‑page report available • Cloudflare Workers free tier provides 100,000 daily requests for server‑less APIs