Claude Opus 5 API: 10% Below List, Alipay/Card, OpenAI-Compatible (2026)
Opus 5 and Sonnet 5 are live on cocodot. Opus 5 costs the same as 4.8 with double the context; Sonnet 5 is officially priced BELOW Sonnet 4.6. Both at 90% of list.
Opus 5: same price, double the context
Opus 5 lists at exactly Opus 4.8's price ($5/$25) with context doubled to 1M and a generational capability bump. There is no reason left to pin 4.8 — same bill, newer model. If your code pins claude-opus-4-8, change it to claude-opus-5; the generic alias "opus" still routes to 4.8 for stability, so use the full name to get 5.
Sonnet 5: the first Claude generation to launch cheaper
Sonnet 5 lists at $2/$10 versus Sonnet 4.6's $3/$15 — a third cheaper. Sonnet is the industry's default coding tier (Cursor, Claude Code, Cline), so this directly cuts everyday development cost. On cocodot Sonnet 5 is $1.8/$9. Note our Sonnet 4.6 runs at 80% of its list ($2.4/$12) — still the cheapest Claude on the menu if you need version pinning; otherwise just move to 5.
Two-minute setup
OpenAI-compatible: base_url https://cocodot.co/api/ai/v1, model claude-opus-5 or claude-sonnet-5 (official names auto-route; the codes mco-7/mcs-6 work too). Claude Code users: ANTHROPIC_BASE_URL=https://cocodot.co/api/ai. Alipay or card, metered billing, $0.5 trial credit on signup (flagships need a small top-up).
How to split your workload
At a 3:1 input:output mix, Sonnet 5 blends to roughly $3.6 per M tokens and Opus 5 to about $9 — a 2.5x gap. Sensible split: completions, refactors and everyday Q&A on Sonnet 5; cross-file rewrites, long agent chains and genuinely stuck problems on Opus 5. Same key, same balance, switch by model name.
Claude 5 generation vs previous: list price and cocodot price (USD per M tokens)
| Model | List (in/out) | cocodot (90%) | Context |
|---|---|---|---|
| Opus 5 | $5 / $25 | $4.5 / $22.5 | 1M |
| Opus 4.8 (prev) | $5 / $25 | $4.5 / $22.5 | 500K |
| Sonnet 5 | $2 / $10 | $1.8 / $9 | 1M |
| Sonnet 4.6 (prev) | $3 / $15 | $2.4 / $12 (80% of list) | 1M |