🚨 Anthropic Drops Claude Sonnet 5 — 1M Context Window, Opus-Level Agentic Coding at 60% Lower Cost
What happened
On June 30, 2026,
Anthropic launched Claude Sonnet 5 — their most agentic Sonnet model yet. This isn't a marginal upgrade. It's a model that can make plans, use tools like browsers and terminals, and run autonomously at a level that, just a few months ago, required their much more expensive Opus-class models.
Key details from the launch:
- 1M-token context window — up from 128K in Sonnet 4.6. You can now load entire codebases, massive documentation sets, or full-length logs into a single prompt without chunking.
- Agentic coding: 63.2% on SWE-bench Pro — Sonnet 4.6 scored 58.1%. Opus 4.8 still leads at 69.2%, but Sonnet 5 closes the gap significantly while costing ~60% less.
- Introductory pricing: $2/MTok input, $10/MTok output through August 31, 2026 (then $3/$15). Compare that to Opus 4.8 at $15/$75 — Sonnet 5 is 87% cheaper on input and 87% cheaper on output.
- New default for Free and Pro plans on claude.ai — every user gets Sonnet 5 by default now.
- Available in Claude Code, Claude Platform API (model ID:
claude-sonnet-5), and on Amazon Bedrock.
- 128K max output tokens — the model can generate very long responses in a single pass.
Why developers should care
(message trimmed for Discord length limit)