claude-opus-5-5 API — per-token pricing and code examples
Price per 1M tokens
Cache rates apply when the provider reports cache tokens.
78% offAbout this model
claude-opus-5-5 is served through OpenAI Chat Completions, OpenAI Responses, Anthropic Messages. You pay $0.88 per 1M input tokens and $4.40 per 1M output tokens. That is 78% below the official price of $4.00 and $20.00. Cache reads cost $0.088 per 1M tokens when the provider reports cached tokens. It is included in budget packs from $99.00 up.
Where Claude Opus 5.5 fits
Claude Opus 5.5 is the Opus model Anthropic released on September 22, 2026, with a retirement date of not sooner than September 22, 2027 and knowledge up to June 2026. Anthropic describes it as built for long-running agentic coding and knowledge work, rates its latency as moderate, and gives a plain recommendation: if you are unsure which model to use, start with Claude Opus 5.5 for most workloads. It reads text and images, writes text, and has a 1M-token context with up to 128K output tokens per synchronous response.
That makes Opus 5.5 the sensible first choice for a coding agent, a research assistant that reads long documents, or any task where you would rather measure a strong default than guess between models.
Always thinking, medium by default
Adaptive thinking on Opus 5.5 is always on and cannot be switched off. Anthropic's wording is direct: control thinking depth with the effort parameter. Sending thinking: {type: "disabled"}, between_tools, or the old budget_tokens form all return a 400 error, so code written for Opus 5 that disables thinking will break here.
The default effort is medium, one level lower than Opus 5 and earlier Opus models, which default to high. Anthropic points out that a request that omits effort therefore runs one level lower than it did on Opus 5. If you migrate and notice shallower answers on hard tasks, set output_config.effort to high or xhigh explicitly. All five levels, low through max, are accepted.
Two more breaking changes from Opus 5 matter for agents. Forced tool use now returns an error; Anthropic's replacement is tool_choice: {"type": "auto"} together with strict tool use or structured outputs. And the earlier computer_20251124 computer use tool is not accepted on the Claude API. Thinking blocks are tied to the model and the conversation, and with the default omitted display, text the model writes between tool calls arrives in thinking blocks whose text is empty.
What it costs on Cent AI
Every Opus 5.5 call spends some output on reasoning, because thinking cannot be disabled. Those reasoning tokens are billed at the output rate from your prepaid USD balance, using the rates in the price box. The medium default is already a reasonable cost setting; raising effort to xhigh or max buys depth at the price of more output per request.
The minimum cacheable prompt on Opus 5.5 is 512 tokens. Anthropic offers a 5-minute cache by default and a 1-hour cache on request, and on Cent AI cache writes for both lifetimes are billed at the input rate. Cent AI only bills cache reads and writes that the model's usage report counts. Keep effort steady across an agent session, because a changed effort level invalidates cached message blocks.
Using it from your tools
Opus 5.5 is available on /v1/messages, /v1/chat/completions and /v1/responses, and Cent AI does not change your parameters silently. Remove temperature, top_p and top_k from requests: Anthropic documents that non-default values fail with 400 on Opus 5.5, and from API version 2026-10-21 any value fails. A prefilled final assistant turn is also rejected. Where you used to force a tool, ask for JSON with output_config.format instead.
A /v1/messages request in the shape of Anthropic's Messages API reference, replacing a forced tool with a JSON schema:
{
"model": "claude-opus-5-5",
"max_tokens": 4096,
"output_config": {
"effort": "high",
"format": {
"type": "json_schema",
"schema": {
"type": "object",
"properties": { "risk": { "type": "string" } },
"required": ["risk"],
"additionalProperties": false
}
}
},
"messages": [{ "role": "user", "content": "Rate the risk of this change" }]
}
The model name on Cent AI is claude-opus-5-5, the same ID Anthropic uses on its own API.
Specification
- Context window
- —
- Input and output
- Text
- Endpoints
- /v1/chat/completions, /v1/responses, /v1/messages
- Smallest pack
- $99.00
Latency and error rate per model are on the status page. status.cent-ai.dev
What a request costs
Open the full calculatorCost per request
Estimate
claude-opus-5-5An estimate at price version 3. Your bill comes from the usage the provider reports for each request.
Save $0.7428Code
curl https://api.cent-ai.dev/v1/chat/completions \
-H "Authorization: Bearer $CENT_AI_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "claude-opus-5-5", "messages": [{"role": "user", "content": "Hello"}]}'Set CENT_AI_KEY to a key from the dashboard. OpenAI SDKs use https://api.cent-ai.dev/v1; Anthropic SDKs use https://api.cent-ai.dev.
Price history
Version 3, in effect since Oct 31, 2026.
Effective
per 1M tokens
Inputold $0.88 new $0.88Cache readold $0.088 new $0.088Cache writeold $1.10 new $0.88Cache write 1hold $1.76 new $0.88Outputold $4.40 new $4.40Effective
per 1M tokens
Inputold $0.64 new $0.88Cache readold $0.064 new $0.088Cache writeold $0.80 new $1.10Cache write 1hold $1.28 new $1.76Outputold $3.20 new $4.40
Compare with
- deepseek-v4.1-flash-uncensoredInput$0.156
- claude-sonnet-5.5Input$0.58
- gpt-6-astraInput$3.20
- claude-opus-5Input$1.15
Questions
How much does claude-opus-5-5 cost on Cent AI?
$0.88 per 1M input tokens and $4.40 per 1M output tokens. Cache reads cost $0.088 and cache writes $0.88 per 1M tokens. Each request is rounded up once.
Is there a minimum charge?
A successful request costs at least $0.00002. A request that ends early, because the client left or the output was cut off, pays only for its tokens.
How do I pay?
Top up a prepaid USD balance with QRIS or USDT/USDC, or buy a budget pack once. Nothing renews automatically.
What if the provider does not report usage?
The request is billed from an estimate and marked estimated on the Requests page. An estimate never contains cache tokens.
Which endpoint do I call?
/v1/chat/completions, /v1/responses, /v1/messages. The OpenAI base URL is https://api.cent-ai.dev/v1.
Try claude-opus-5-5 with one key
One prepaid balance for every model. Pay with QRIS or USDT/USDC.