Claude Sonnet 5 ships as agentic model; native 1M context, $2/$10 per Mtok promotional pricing
Anthropic released Claude Sonnet 5 on June 30, 2026, positioning it as the company's most agentic model in the Sonnet tier. Sonnet 5 features substantial improvements over Sonnet 4.6 in reasoning, tool use, coding, and knowledge work. It ships with a native 1-million-token context window and promotional pricing of $2 per million input tokens and $10 per million output tokens through August 31, 2026 (after which standard pricing resumes).
Sonnet 5 is optimized for multi-step agentic workflows and is already the default model in Claude Code, Anthropic's coding agent. The model is available across all major deployment channels: Anthropic's API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. This release continues Anthropic's accelerated cadence: Sonnet 5 follows Claude Opus 4.8 (May 2026), Claude Fable 5 (June 2026, after export-control suspension and redeployment July 1), and Claude Mythos 5.
The Mythos-class models (Fable 5, Mythos 5) remain above Sonnet in capability, but Sonnet 5 fills the gap for cost-sensitive production agentic use. On the Epoch Capabilities Index as of June 2026, Claude Fable 5 scored 161 points, beating GPT-5.5 Pro by 1 point. Sonnet 5's exact benchmark scores relative to GPT-5.5 are not yet publicly released in detail.
For practitioners, Sonnet 5 at promotional pricing offers a substantive cost reduction for long-context agentic work—the 1M context and agentic focus are material shifts from Sonnet 4.6. Architectural improvements (hybrid reasoning, tool use) suggest this is a tuning/capability jump rather than infrastructure revolution. The promotional period through end-August creates a time-window incentive for enterprises to lock in discounts on high-volume agentic deployments.