Claude Code's $2/$10 introductory API pricing runs out on 31 August
Sonnet 5 is the default model in Claude Code, and its API has been running on introductory rates that expire at month end. GitHub Copilot has already moved every plan to usage-based billing with a monthly credit allocation.
Introductory pricing has a way of setting expectations it was never meant to set. Teams have spent months building agent workflows whose unit economics assume $2 per million input tokens, and the budget conversation on 1 September will be about a number nobody has published yet.
The direction across the category is consistent. Copilot moved all plans to usage-based billing with an included monthly allocation and overage at month end; Cursor split its usage pools and added a power seat. The flat-rate era in coding assistants is closing because agentic usage does not fit a flat rate.
What survives is multi-tool stacks. Most teams run an autocomplete tool and an agentic one, and metered pricing on both makes the routing decision — which model gets which task — a cost question rather than a preference.
Spectrum AI Lab — AI Coding Tools Pricing 2026: Copilot vs Claude Code vs Codex vs Cursor → · Developers Digest — Best AI Coding Tools July 2026: Updated After Opus 5 and Fable 5 API-Only →