Anthropic released Claude Sonnet 5.5 on September 28, 2026, and it is now available on GetVibe API as claude-sonnet-5.5. Same key, same endpoint, same per-token billing from your USD balance. Only the model value changes.
What Anthropic changed
Anthropic keeps the Sonnet 5 list price and says Sonnet 5.5 runs more than 30% faster and costs up to 30% less on most work, because it needs fewer tokens per task. The biggest jump is in agentic work. From Anthropic's announcement:
| Benchmark | Sonnet 5.5 | Sonnet 5 | Opus 5.5 |
|---|---|---|---|
| Terminal-Bench 4.0 | 70.6% | 10.3% | 66.4% |
| CursorBench 4.0 | 55.5% | 34.1% | 57.8% |
| OSWorld 2.1 | 80.1% | 57.0% | 81.8% |
| Humanity's Last Exam (with tools) | 64.5% | 54.9% | 67.7% |
| GDPval-AA v2.1 (Elo) | 1844 | 1449 | 1846 |
These are vendor numbers. Check them on your own workload before you switch a production agent.
Specs, and what we measured before launch
| Claude Sonnet 5.5 | |
|---|---|
| Model ID | claude-sonnet-5.5 |
| Context window | 1,000,000 tokens |
| Max output | 128,000 tokens |
| Input | text and images |
| Tool calling | yes (tool_choice: auto) |
| Price | $0.0534 input · $0.2667 output per 1M tokens |
A 270,000-token prompt came back with the markers we planted at both its start and its end, and an image test read the number and both colours correctly. Two request parameters behave differently from Sonnet 5:
- Forced tool choice is not supported. The model rejects
tool_choice: "required"or a named function. GetVibe turns such a request intoautowith the same tools, so you get an answer instead of an error, but the model decides whether to call the tool. Name the tool in your prompt, and use structured outputs when you need strict JSON. top_pis deprecated. GetVibe drops it and tells you in a response header.temperatureworks as before.
Call it
curl https://getvibeapi.com/v1/chat/completions \
-H "Authorization: Bearer $GETVIBE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-5.5",
"messages": [{"role": "user", "content": "Review this diff and list the risky changes: ..."}]
}'Tools that speak the Anthropic Messages format (the Anthropic SDK, Claude Code) use /v1/messages on the same host. See the Claude Code guide.
When to pick it
- Coding agents and terminal work, where the gain over Sonnet 5 is largest.
- Everyday development and automation, where Claude Opus 5.5 costs more than the task needs.
- If you already run Sonnet 5, change the model name and check only
tool_choiceandtop_p.
Long agent loops repeat the same prefix, so prompt caching cuts the bill further: cached reads cost about a tenth of the input rate. Live prices are on the model page.