Anthropic Dropped Claude Sonnet 5, a Near-Flagship Model Priced Like a Clearance Sale That Expires in August
2026-06-30
“Anthropic built a model almost as good as its flagship, priced it like a closing-down sale, and set the discount to run out right after you are hooked.”
Anthropic shipped Claude Sonnet 5, its most agentic Sonnet yet, with a 1M token context window and quality that lands close to its pricier Opus 4.8 flagship. It is already the new default for Free and Pro users on claude.ai and live everywhere developers actually work: Claude Code, the API, Cursor, VS Code and GitHub Copilot. As a model, it is a genuinely strong, genuinely useful release. Sally has notes anyway.
Because the headline is the price, and the price has a clock on it. Introductory rates of 2 dollars in and 10 dollars out per million tokens run only through August 31, after which it jumps to 3 and 15. That is a 50 percent hike scheduled the moment your agents are wired into it. Cheap now, pricier later, dropped right as Anthropic sprints toward a blockbuster IPO, is the oldest on-ramp in software wearing a new coat.
The other tell is the framing: a cheaper way to win the agentic race. That is the honest admission that Opus still owns the very hardest problems, with Sonnet 5 trailing it on the toughest coding benchmarks. So this is not the best model Anthropic makes, it is the one priced to take the volume. Smart business. Just remember that a discount with an expiry date is a trial that bills you later.
- Anthropic launched Claude Sonnet 5 (model id claude-sonnet-5) on June 30, 2026, billed as its most agentic Sonnet yet, with a 1M token context window.
- Introductory pricing is 2 dollars in and 10 dollars out per million tokens through August 31, 2026, then rises to 3 and 15; Opus 4.8 costs 5 and 25.
- It is the new default for Free and Pro users on claude.ai and is live in Claude Code, the Claude API, Cursor, VS Code and GitHub Copilot.
- On the SWE-bench Pro agentic coding benchmark it scores 63.2 percent, up from Sonnet 4.6's 58.1 percent, with Opus 4.8 still ahead at 69.2 percent.
- It slightly outperforms Opus 4.8 on a knowledge-work benchmark and scores 84.7 percent on the BrowseComp 25 web-search benchmark, as Anthropic races toward a blockbuster IPO.
- 01
This is real value, not just spin: near-flagship agentic performance at a fraction of the cost, a huge 1M token context window, and immediate availability inside the exact tools developers already use. For most agent workloads, expiry date or not, Sonnet 5 is the new sensible pick.
- 01
Anthropic's own pricier Opus, quietly undercut by its little sibling, and every rival lab now dragged into an agentic price war they did not choose. Plus any developer who wires agents onto the 2 dollar intro rate and forgets it climbs 50 percent on September 1.
No meter. No surprises.
The story isn't Copilot. It's the meter. Here is the calmer way to get your code reviewed.
- Flat price for the Full Suite. No usage meter, no month-end surprise.
- Free CLI: 90 roasts a month, no account needed.
- Privacy-first: a dry-run shows the exact payload, and your secrets never leave your machine.
Works in Claude Code, Cursor and Windsurf via MCP. Open source, and proud of it.
Got something the world should see roasted? Drop it.
A full teardown from €2,99. No mercy.