Claude Sonnet 5 API: The Default Coding Tier Just Got Cheaper (2026)
Sonnet is what Cursor, Claude Code and Cline actually run all day. Sonnet 5 lists at $2/$10 — a third below Sonnet 4.6 — with 1M context. cocodot serves it at $1.8/$9, Alipay or card, metered.
Why Sonnet is the tier that matters
Check any coding tool's defaults: Cursor defaults to Sonnet, Claude Code's main model is Sonnet, Cline recommends Sonnet. Completions, refactors and everyday Q&A are 80% of development calls, and Sonnet's quality is sufficient at a fraction of Opus cost. Flagships are for the hard 20%; the bill lives in the default tier.
What makes this launch unusual
Every previous Claude upgrade cost more. Sonnet 5 reverses that: list price dropped from $3/$15 to $2/$10 while the model got a generational bump. Migrating from mcs-5 to mcs-6 is a one-string change that cuts the same workload's bill by 25% — there is no catch to weigh.
Setup in two minutes
OpenAI-compatible: base_url https://cocodot.co/api/ai/v1, model claude-sonnet-5. Claude Code: export ANTHROPIC_BASE_URL=https://cocodot.co/api/ai and ANTHROPIC_MODEL=mcs-6. Pay by Alipay or card, metered per token; email signup grants $0.5 to verify cheap tiers first.
When to stay on 4.6
Two cases: production pipelines with version-pinned regression baselines, and maximum price sensitivity — our 4.6 carries the deepest discount on the menu (80% of list). Everyone else should simply move: newer and cheaper is not a trade-off.
Two Sonnet generations (cocodot pricing, USD per M tokens)
| Model | Code | List | cocodot | Context |
|---|---|---|---|---|
| Sonnet 5 | mcs-6 | $2 / $10 | $1.8 / $9 (90%) | 1M |
| Sonnet 4.6 | mcs-5 | $3 / $15 | $2.4 / $12 (80% — our deepest discount) | 1M |