Claude Sonnet 5: 60% off, then 40% off Opus 4.8

Anthropic released Claude Sonnet 5 on June 30, 2026, with an introductory pricing tier of $2 per million input tokens and $10 per million output tokens, lasting until August 31, 2026 [DevTo]. After that date, the price moves to $3/$15, still 40% cheaper than the Opus 4.8 baseline of $5/$25 [Anthropic Blog]. Sonnet 5 is listed across every Claude tier and was added to Copilot on launch day, making it the default for code-completion in the IDE [GitHub Blog].

A typical coding-agent pipeline consuming 50M input and 10M output tokens per month would cost $200 under the intro Sonnet 5 rate versus $500 for Opus 4.8—a $300 saving [DevTo]. Even after the discount ends, the same workload drops to $300 versus $500, still a $200 reduction. Broad availability in Copilot and the Claude API lowers integration friction, allowing teams to switch without rewriting tooling. Competing models like GPT-5.6 and Gemini 3.5 Pro remain gated or expensive, limiting their production use this quarter.

The post-August price still leaves a 40% margin, making Sonnet 5 a cost-effective option for routine agent work [Anthropic Blog]. Companies that lock in Sonnet 5 for the bulk of their agent traffic will see measurable savings this quarter. The discipline of model routing, not the model itself, is key to reducing costs.