Daily Brief · 23 September 2026
Anthropic ships Claude Opus 5.5: Fable-level work at ~40% less cost
Anthropic released Claude Opus 5.5 on Tuesday—the first model in its Claude 5.5 family and its first launch since CEO Dario Amodei’s call to pace the frontier. The company says Opus 5.5 matches Claude Fable 5.1 on most work while costing about 40% less to run than Opus 5 on typical workloads, with API pricing at $4 / $20 per million input/output tokens (down from Opus 5’s higher list). External evaluators including Frontier Design and METR tested it pre-release; Anthropic claims its strongest automated behavioral-audit scores to date, plus preserved-thinking anti-distillation for newer API accounts. Availability spans Claude apps, the Claude API (claude-opus-5-5), Bedrock, Google Cloud, and Microsoft Foundry. Sonnet 5.5 and Haiku 5.5 are promised in the coming weeks. Implication: frontier labs are competing on token efficiency and price as much as peak capability.
What else matters
OpenAI cuts GPT-6 Sol and Luna API prices 50%
OpenAI expanded the GPT-6 lineup with Sol and Luna—faster, cheaper tiers built on Astra’s coding, factuality, computer-use, and alignment work. API list pricing is half GPT-5.6 promotional rates: Sol at $2 / $10 per million input/output tokens for complex coding and agentic workflows; Luna at $0.10 / $0.50 for high-volume tasks. Both ship in the API (gpt-6-sol, gpt-6-luna); ChatGPT Work/Codex cover paid tiers, with Luna also for Free/Go on desktop. Same-day as Opus 5.5—clear pressure from cheaper open-weight competitors.
OpenAI Developer Community announcement · Sol docs · Luna docs
Snorkel AI raises $350M at $3.5B valuation
Data-labeling and evaluation startup Snorkel AI closed $350 million in fresh funding at a $3.5 billion valuation, CEO Alex Ratner told Reuters, citing surging demand for complex AI training data from frontier labs. The round underscores that high-quality, specialist data—not just compute—remains a scarce input as model makers chase agentic and domain workloads.
Reuters via NewsX — Snorkel AI
Ex-Google safety chief joins Common Sense Youth AI Safety Institute
James Siegel, who founded Google’s trust-and-safety team, warned that AI could harm children more than social media did and called for a development slowdown. On Tuesday he joined Common Sense Media as first executive director of its Youth AI Safety Institute, pushing age verification at chatbot providers and independent “crash-test” style standards for consumer AI tools.
