Claude Sonnet 5.5 Is Live on Felo OpenAPI

Anthropic's newest Sonnet lands on Felo OpenAPI on launch day. Same $2/$10 pricing, 1M context, 30% faster — and one API key for Felo Pro users.

Anthropic shipped Claude Sonnet 5.5 on September 28, 2026. You can call it today.

Felo OpenAPI added the claude-sonnet-5-5 route on launch day, at the same $2 per million input tokens and $10 per million output tokens that Anthropic charges — no waiting for a reseller to catch up, no separate account, no new SDK to learn.

Try Claude Sonnet 5.5 on Felo OpenAPI: https://openapi.felo.ai/models/anthropic/claude-sonnet-5.5

Why this release matters

Sonnet 5.5 is not a new flagship. It is the model most teams actually run in production, and Anthropic spent this release making it cheaper to run at scale.

The pitch is a straight trade: more speed, lower cost, same intelligence class. Anthropic positions Sonnet 5.5 as the best combination of speed and intelligence in its lineup, and the numbers back the framing.

	Claude Sonnet 5	Claude Sonnet 5.5
Input / output	$2 / $10 per MTok	$2 / $10 per MTok
Context window	1M tokens	1M tokens
Max output	128K tokens	128K tokens
Latency	Balanced	Fast
Thinking	Adaptive	Adaptive, on by default
Default effort	—	High
Knowledge cutoff	—	June 2026
Released	June 30, 2026	September 28, 2026

Same price, same context, faster route. Anthropic's own guidance is blunt: Sonnet 5 still works, but you should migrate to Sonnet 5.5.

What actually changed under the hood

Three things matter for anyone running agents in production.

Adaptive thinking is now the default. Sonnet 5.5 decides how much to think based on the task instead of requiring you to budget tokens up front. The effort parameter controls the depth, and the default is high. You get reasoning quality without hand-tuning a thinking budget per request.

The tokenizer is unchanged. Anthropic kept the same tokenizer as Sonnet 5, so the same text produces the same token counts. If you already modeled your costs against Sonnet 5, your estimates still hold. That is a bigger deal than it sounds — a tokenizer change quietly invalidates every cost projection and cache assumption you have.

Cache reads got cheaper. Prompt caching is where agent loops actually spend money, and Sonnet 5.5 keeps cache reads at $0.20 per million tokens while running faster. Long agent sessions that replay the same context get cheaper per turn.

Five breaking changes before you migrate

Sonnet 5.5 is a drop-in upgrade for most workloads, but Anthropic flagged five breaking changes that affect code already running on Sonnet 5. If you have a working agent, read these before you switch the model ID.

1. thinking: {"type": "disabled"} now returns a 400 error. Use thinking: {"type": "between_tools"} instead. It is the lowest thinking setting on this model and needs no beta header.
2. Forced tool use is gone. tool_choice set to {"type": "any"} or {"type": "tool", "name": "..."} returns a 400. auto and none still work. For schema-valid input, keep tool_choice: {"type": "auto"} and set strict: true.
3. Thinking blocks are tied to the model that produced them. Sonnet 5.5 reads thinking blocks from Sonnet 5, Opus 4.8, Haiku 4.5, and earlier models — but not from Opus 5, Opus 5.5, or any Fable or Mythos model. A conversation moving from Sonnet 5 to Sonnet 5.5 keeps its reasoning; moving off Sonnet 5.5 does not carry it forward.
4. The older computer_20251124 computer-use tool is not accepted on the Claude API and Google Cloud.
5. The advisor tool rejects Opus 4.8, Opus 4.7, and Sonnet 5 as advisors.

There is also one silent change: text between tool calls now comes back inside thinking blocks. If your app streams that text to users, it will go quiet between tool calls until you set a display value to return it, or turn off up-front thinking with between_tools. No error, just a stream that stops talking.

The full migration guide is on Anthropic's docs.

Calling Sonnet 5.5 on Felo OpenAPI

The route is live. Here is the whole integration.

curl https://openapi.felo.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $FELO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5-5",
    "messages": [
      {"role": "user", "content": "Summarize this changelog and flag breaking changes."}
    ]
  }'

Felo OpenAPI exposes three compatible surfaces, so you can keep the client you already use:

Surface	Method	URL
Chat Completions	POST	https://openapi.felo.ai/api/v1/chat/completions
Responses	POST	https://openapi.felo.ai/api/v1/responses
Messages (Anthropic-compatible)	POST	https://openapi.felo.ai/api/v1/messages

Because the Messages endpoint is Anthropic-compatible, Claude-style clients only need the base URL pointed at https://openapi.felo.ai/api. Third-party SDKs work the same way — set baseURL to https://openapi.felo.ai/api/v1 and keep your existing code.

Add "stream": true for server-sent events. Sonnet 5.5 supports streaming, tool calling, structured output, and vision on the same route.

If you already pay for Felo Pro, you already have access

This is the part worth repeating: Felo Pro includes API access to Felo OpenAPI models.

If you subscribe to Felo Pro, you do not need a second subscription to call Sonnet 5.5. Create an API key from your Felo account and point your agent at the endpoint. Your existing Felo credits cover model calls, and the same key works across the model lineup — Sonnet 5.5, Opus 5.5, GPT-6, Grok 4.7, and the rest.

Two things to keep straight:

• Felo Pro credits are one pool. Pro Search, research agents, slides, images, and API model calls all draw from the same balance. Model calls bill by token, so heavy API use will move through credits faster than a few searches a day.
• Tool APIs are separate. Model calls are covered by your Pro access. The harness tools — Web Fetch, X Search, PPT generation, LiveDoc, YouTube subtitling — run on Felo API Platform credits.

For most Felo Pro users, the practical setup is: keep using Felo for search and creation, and add an API key when you want an agent like Claude Code or Codex to call Sonnet 5.5 directly. One account, one key, one bill.

Not on Pro yet? Every Felo account gets 200 free credits per day, and the free tier can test selected models and tools before you commit. It is enough to run real requests against Sonnet 5.5 and see how it behaves on your own prompts.

Where Sonnet 5.5 fits

Sonnet 5.5 is the route to pick when you want near-flagship quality without flagship pricing.

• Coding agents — long-running edits across a codebase, where the 1M context window holds the files and the fast route keeps the loop responsive.
• Document and research workflows — 1M tokens of context for long reports, contracts, or transcripts, with structured output for downstream parsing.
• High-volume production calls — classification, extraction, and summarization where per-token cost decides whether the feature ships.
• Multilingual work — Anthropic calls out multilingual tasks as a Sonnet 5.5 strength, which pairs well with Felo's own multilingual focus.

If you need the absolute ceiling on hard reasoning, Opus 5.5 is still the flagship and it is on Felo OpenAPI too. Sonnet 5.5 is the model you run when the work is real but the budget is not unlimited.

Try it now

Sonnet 5.5 is live on Felo OpenAPI today, at Anthropic's list price, behind the key you already have.

Start in the Playground to test prompts without writing code, then grab an API key and point your agent at the endpoint. If you are on Felo Pro, you are one key away from running Anthropic's newest Sonnet in your own stack.

→ Try Claude Sonnet 5.5 on Felo OpenAPI

→ Create your API key

---

This post is also available in 简体中文, 日本語, 한국어, 繁體中文, हिन्दी, Français, العربية, Русский, اردو, Bahasa Indonesia, Deutsch, Tiếng Việt, Türkçe, Italiano, ไทย, Español, বাংলা and Português.