Claude Sonnet 5's $2/$10 Pricing Ends August 31, Rises 50%

Anthropic released Claude Sonnet 5 on June 30, 2026, at an introductory price of $2 per million input tokens and $10 per million output tokens. That introductory rate holds only through August 31, 2026, and on September 1 it rises to $3 and $15, a 50% jump on both sides. If you’re sizing a budget on today’s rate, build the September number in now, because the bill changes on a fixed date, not a usage threshold.

The Deadline That Matters: August 31

Anthropic calls it “introductory pricing,” and it’s dated explicitly in the release: $2/$10 per million tokens through August 31, 2026, then $3/$15 after. My take: that’s not a small print detail, it’s the actual headline for anyone running Sonnet 5 at volume. A workload that costs $2,000 a month in output tokens today costs $3,000 a month starting September 1, with no code change on your end.

The Most Agentic Sonnet Yet

Anthropic describes Sonnet 5 as “the most agentic Sonnet model yet. It can make plans, use tools like browsers and terminals, and run autonomously,” at a level the company says used to require a larger, more expensive model. Anthropic also says its performance is “close to that of Opus 4.8, but at lower prices,” and that it’s a substantial jump over Sonnet 4.6 on reasoning, tool use, coding, and knowledge work specifically.

Anthropic hasn’t published a context window figure for Sonnet 5. Don’t carry over the number from an older Sonnet release; treat it as unstated until Anthropic says otherwise.

The release also carries a safety comparison worth reading if you plan to run this model unattended. Anthropic says its assessments found Sonnet 5 shows “an overall lower rate of undesirable behaviors than Sonnet 4.6,” and is “generally safer to use in agentic contexts.” It adds that Sonnet 5 has “a much lower ability to perform cybersecurity tasks than our current Opus models,” a distinction that matters when you are picking which model gets to touch real systems.

What Early Access Partners Are Seeing

Anthropic’s release quotes several early access partners on what changed in practice. At Lovable, the team said Sonnet 5 “gets more done with less,” same output quality in fewer steps, and “refuses unsafe requests cleanly and consistently.” At Pace, where computer-use agents run insurance workflows like submission intake and loss runs, the team said Sonnet 5 “consistently takes the right action and does it quickly, which is what real insurance work demands.” At ClickHouse, the team said Sonnet 5 “reasons in tighter steps and gets our users to answers noticeably faster.”

Pricing and Availability

Sonnet 5 is the default model for Free and Pro plans, and it’s available to Max, Team, and Enterprise users, in Claude Code, and on the Claude Platform. Developers reach it as claude-sonnet-5 via the Claude API. Pricing: $2/$10 per million input/output tokens through August 31, 2026, then $3/$15 after.

What It Means for Developers

Sonnet has historically been the model teams default to for volume work, and Anthropic is explicit that Sonnet 5 narrows the gap to Opus-class agentic performance while staying cheaper. If you’re evaluating it now, the honest comparison is against September pricing, not the launch rate you’ll actually be paying for most of the model’s life. Pair that with Claude Opus 5, released a few weeks later at an unchanged $5/$25, and Anthropic’s current lineup gives you a real choice between a cheaper agentic default and a flagship whose price hasn’t moved since Opus 4.5.