Claude Sonnet 4.6: Legacy Model Reference

Claude Sonnet 4.6 was Anthropic's prior general-purpose default. Sonnet 5 now fills that role on Claude Free and Pro and is available in Claude Code and the API. Keep this page as a legacy reference for teams maintaining Sonnet 4.6 evaluations, prompts, or pinned integrations; use Sonnet 5 as the starting point for new work.

What Claude Sonnet 4.6 is best for

Sonnet 4.6 is the strongest agentic coding model in its class and pairs with Claude Code, Anthropic's dedicated CLI. The 1M-token context window lets it reason over large codebases and big document sets in a single pass, which makes it a natural fit for retrieval-augmented generation and production assistants.

• Agentic coding with Claude Code as the dedicated CLI
• Long-context analysis over large codebases and document sets
• Production assistants and RAG pipelines
• Tool-using agents, with first-class Model Context Protocol support

Pricing

Claude Sonnet 4.6 is priced at $3 per million input tokens and $15 per million output tokens. Batch processing cuts costs by 50%, and prompt caching reduces cached input by up to 90%. Compared with Opus 4.8 at $5 in and $25 out, Sonnet 4.6 is the clear value pick for high-volume work.

Context window

Sonnet 4.6 ships with a 1M-token context window at flat rates, with no long-context surcharge. That matches the Opus 4.8 flagship and far exceeds the 200K window on Haiku 4.5. You can load entire repositories or large reference sets without paying a premium for the larger window.

How it compares

Sonnet 4.6 is a legacy benchmark for prior integrations. Sonnet 5 is now the default on Claude Free and Pro and is available in Claude Code and the API. Fable 5 was restored globally on July 1, 2026. Mythos 5 is limited to approved Project Glasswing organizations. For a new model decision, compare current options against your own production workload rather than relying on this older tiering.

Should you use it

If you want a single default that handles coding, long-context analysis, and tool-using agents well, the answer is yes. Reach for Opus 4.8 only when you need the absolute top of agentic coding or very long autonomous runs. Note the trade-offs that apply across the lineup: there is no native image generation, the focus is text and code, and voice exists in the mobile apps but is not a core differentiator.

THE CLAUDE INSIDER

Get the Claude playbook in your inbox.

One weekly email for Claude and Claude Code users. Real workflows, no hype. Subscribe and we send you The Claude Power-User Cheatsheet.