Anthropic unveils Claude Sonnet 5 AI model as US blocks Fable 5

Anthropic introduced Claude Sonnet 5, its latest large‑language model, following the withdrawal of the more powerful Fable 5 after the U.S. government barred the model from non‑American users citing national‑security concerns. Sonnet 5 is positioned just below the costlier Opus 4.8 in benchmark performance, matching it on tasks such as coding, reasoning and tool use, while offering lower token pricing — $2 per million input tokens and $10 per million output tokens until 31 August, rising thereafter. The company highlights added safety guardrails, reduced hallucinations and stronger resistance to prompt‑injection attacks compared with its predecessor Sonnet 4.6, though internal audits still show a higher share of misaligned behaviour than the more capable Opus 4.8. The model is immediately available to free, Pro, Max, Team and Enterprise users via Claude Code and the Claude platform. Independent tests confirm its claims, and the rollout coincides with OpenAI’s limited launch of GPT‑5.6 variants, underscoring a shifting competitive landscape in generative AI.