Claude Sonnet 5 Pricing: the $2/$10 Intro Rate Is Now Standard
Anthropic made Sonnet 5's $2/$10 introductory rate its standard price and cancelled the rise to $3/$15 scheduled for September 1, 2026. What that settles about routing and cost planning.
Fable5 EditorialClaude Sonnet 5 (claude-sonnet-5) lists at $2 / $10 per million input/output tokens, and that is no longer a promotion: Anthropic has made the launch-time introductory rate its standard price and cancelled the step-up to $3 / $15 that was scheduled for September 1, 2026 (checked 2026-09-19). Here is what that settles, and how Sonnet 5 sits against Opus 5 and Fable 5.
What Sonnet 5 costs
Sonnet 5 is Anthropic’s mid-tier, agentic model: a 1M-token context window, up to 128K output tokens, and quality that lands close to Opus 4.8 for a fraction of the price. At the moment it lists at:
- $2 per million input tokens
- $10 per million output tokens
- $2 / $10 is the standard price — the scheduled September 1 rise to $3 / $15 was cancelled (Anthropic pricing page, checked 2026-09-19)
An earlier version of this page carried an introPriceUntil: '2026-10-31' marker in the site’s model reference and argued from it. That date is not on Anthropic’s pricing page, the field has been removed, and the reasoning built on it no longer holds: $2/$10 is the steady state.
Why it matters
At $2/$10, Sonnet 5 is the cheapest current flagship-tier Claude model by a wide margin. That makes it the natural default for work that is:
- High-volume — summarization, extraction, chat, and routine agentic coding where per-token cost dominates.
- Token- or latency-sensitive — fast enough to keep interactive products snappy, at a rate that scales with volume.
The practical consequence is that the rate is now stable. Teams building on Sonnet 5 at $2/$10 no longer need a contingency for a September re-price — which matters because the model is also the cheapest way to buy a 1M-token window in this lineup.
How it compares
| Model | Price (in / out per M tokens) | Context | Best for |
|---|---|---|---|
| Claude Sonnet 5 | $2 / $10 | 1M | Budget, high-volume, speed |
| Claude Opus 5 | $5 / $25 | 1M | Heavy reasoning, default on Max |
| Claude Fable 5 | $10 / $50 | 1M | Long-horizon, security, full-repo |
A simple routing rule: default to Opus 5 at high effort when quality matters most, Sonnet 5 for budget and volume, and Fable 5 only for long-horizon, security-sensitive, or 1M-full-repo escalation where peak reasoning justifies the 5x premium over Sonnet 5. See the full breakdown on the Claude Fable 5 vs Opus 5 and Sonnet 5 vs Fable 5 pages.
What this changes for planning
If you are price-sensitive, the reason to evaluate Sonnet 5 is now structural rather than calendar-driven:
- Run trials on Sonnet 5 now for the workloads you expect to move to it at scale.
- Measure the real cost — including the ~30% tokenizer inflation on old token counts — with the cost calculator.
- Stop discounting the rate for risk of a rise. Budget at $2/$10 as a durable number; the tokenizer inflation, not the list price, is the variable left to manage.
Summary
- Sonnet 5 lists at $2 / $10 per million tokens (input/output).
- That is the standard price, not a promotion — the scheduled $3 / $15 increase was cancelled.
- 1M context, 128K output, and the cheapest flagship-tier rate in the lineup.
- Default: Opus 5 at high effort, Sonnet 5 for budget and volume, Fable 5 for long-horizon escalation.
- Verify the live rate on Anthropic’s pricing page and the Sonnet 5 model page.
Independent analysis. Pricing re-verified against Anthropic’s pricing page on September 19, 2026, when the introductory rate was confirmed as standard — official sources always win on conflict.