claude-opus-4-6 API — per-token pricing and code examples

Checking statusPrice version 5 · in effect since Nov 19, 2026

Price per 1M tokens

InputOfficial price $5.00$1.00
Cache readOfficial price $0.50$0.10
Cache writeOfficial price $6.25$1.25
Cache write 1 hOfficial price $10.00$2.00
OutputOfficial price $25.00$5.00

Minimum per request$0.00002

Cache rates apply when the provider reports cache tokens.

80% off

About this model

claude-opus-4-6 is served through OpenAI Chat Completions, OpenAI Responses, Anthropic Messages. You pay $1.00 per 1M input tokens and $5.00 per 1M output tokens. That is 80% below the official price of $5.00 and $25.00. Cache reads cost $0.10 per 1M tokens when the provider reports cached tokens. It is included in budget packs from $9.99 up.

Where Claude Opus 4.6 fits

Claude Opus 4.6 is an older Opus model from Anthropic that is still served for existing integrations. Anthropic marks it as legacy: it was released on February 5, 2026, and its retirement date is listed as not sooner than February 5, 2027. Its training data reaches May 2025. The model reads text and images and answers in text, with a context window of 1M tokens and up to 128K output tokens per synchronous response.

Anthropic's own model page does not pitch Opus 4.6 for new work and instead suggests moving to Claude Opus 5.5. Keep using Opus 4.6 when you already have prompts and evals tuned for it and need time before you migrate, or when you want an Opus model whose thinking stays off unless you ask for it.

Thinking is off until you turn it on

This is the main difference from the newer Claude models. On Opus 4.6, a request without a thinking field runs with no thinking at all. Anthropic's thinking guide says thinking stays off until you set thinking: {type: "adaptive"}. The older enabled mode with budget_tokens is still accepted, but Anthropic lists it as deprecated, so new code should use adaptive. Sending between_tools returns a 400 error on this model.

When thinking runs, the default display is summarized, so you get a readable summary of the reasoning instead of an empty block. Depth is set with output_config.effort. The default is high, and Opus 4.6 accepts low, medium, high and max. The xhigh level that newer Claude models have is not available here.

Forced tool use behaves differently depending on the thinking mode. Anthropic notes that forcing a tool is incompatible with manual extended thinking but works with adaptive thinking, which is one more reason to drop budget_tokens.

What it costs on Cent AI

Opus 4.6 is charged per token from your prepaid USD balance, using the rates in the price box on this page. Because thinking is off by default, a plain request only pays for the input and the visible answer. Once you switch adaptive thinking on, the reasoning tokens are counted as output, so higher effort makes each answer more expensive even when the final text stays short.

Anthropic distinguishes a 5-minute and a 1-hour cache lifetime, and on Cent AI cache writes for both lifetimes are billed at the input rate. Cache reads and writes are billed only from the token counts the model reports in its usage, never from a guess. On Opus 4.6 the minimum cacheable prompt is 4,096 tokens, which is higher than on the 5.x models. Prompts below that length are processed without caching and without an error, so small system prompts never reach the cache rates.

Using it from your tools

Opus 4.6 works on /v1/messages, /v1/chat/completions and /v1/responses. Cent AI does not change your parameters silently, so Anthropic's rules for this model decide what is accepted. Two rules catch older code. A prefilled final assistant turn is rejected with 400. And under API version 2026-10-21, any temperature, top_p or top_k value is refused; under older API versions only the combination with thinking was restricted.

The example below is for /v1/messages and turns adaptive thinking on at medium effort:

{
  "model": "claude-opus-4-6",
  "max_tokens": 4096,
  "thinking": { "type": "adaptive" },
  "output_config": { "effort": "medium" },
  "messages": [{ "role": "user", "content": "Review this migration plan" }]
}

Example: the cost of one request

Take a request that sends 2,000 new input tokens, reads 8,000 tokens from cache and gets 800 output tokens back. At the rates in the price box on this page:

  • input: 2,000 × $0.5 / 1,000,000 = $0.001
  • cache read: 8,000 × $0.05 / 1,000,000 = $0.0004
  • output: 800 × $2.5 / 1,000,000 = $0.002

That request costs $0.0034, about Rp62 at Cent AI's exchange rate on 2026-11-14 (Rp18,231.96 per USD). A thousand requests like it cost $3.4. Your real requests will differ; each one shows its own amount in usage.cost and on the Requests page.

Questions

Is there a 5-hour or weekly limit? No. You pay per token from your balance. Each key accepts up to 600 requests per minute, and you can set your own daily or monthly limit on a key.

Is a failed request charged? A request that fails before the model receives it costs $0. Tokens the model already produced are charged.

Does unused balance expire? No, not while your account exists. Every top-up is manual, with no auto-renewal.

Compare with claude-opus-5, claude-opus-5-5, claude-sonnet-4-6, or see every model on the pricing page.

Specification

Context window
—
Input and output
Text
Endpoints
/v1/chat/completions, /v1/responses, /v1/messages
Smallest pack
$9.99

Latency and error rate per model are on the status page. status.cent-ai.dev

What a request costs

Open the full calculator

Cost per request

90% is an assumption you can change, not a promise. Cache is billed only from cache tokens the model provider reports.

See how many such requests it covers.

Estimate

claude-opus-4-6
Fresh input99,000 tok
Cache read891,000 tok
Output10,000 tok

Per request$0.2381
At official pricesAt official prices $1.1905

Total$0.2381

An estimate at price version 5. Your bill comes from the usage the provider reports for each request.

Save $0.9524

Code

curl https://api.cent-ai.dev/v1/chat/completions \
  -H "Authorization: Bearer $CENT_AI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "claude-opus-4-6", "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 5, in effect since Nov 19, 2026.

  1. Effective

    per 1M tokens

    Inputold $0.50 new $1.00
    Cache readold $0.05 new $0.10
    Cache writeold $0.50 new $1.25
    Cache write 1hold $0.50 new $2.00
    Outputold $2.50 new $5.00
  2. Effective

    per 1M tokens

    Inputold $0.50 new $0.50
    Cache readold $0.05 new $0.05
    Cache writeold $0.625 new $0.50
    Cache write 1hold $1.00 new $0.50
    Outputold $2.50 new $2.50
  3. Effective

    per 1M tokens

    Inputold $0.25 new $0.50
    Cache readold $0.025 new $0.05
    Cache writeold $0.25 new $0.625
    Cache write 1hold $0.25 new $1.00
    Outputold $1.25 new $2.50
  4. Effective

    per 1M tokens

    Inputold $0.25 new $0.25
    Cache readold $0.025 new $0.025
    Cache writeold $0.3125 new $0.25
    Cache write 1hold $0.50 new $0.25
    Outputold $1.25 new $1.25

Compare with

All models against official prices

Questions

How much does claude-opus-4-6 cost on Cent AI?

$1.00 per 1M input tokens and $5.00 per 1M output tokens. Cache reads cost $0.10 and cache writes $1.25 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-4-6 with one key

One prepaid balance for every model. Pay with QRIS or USDT/USDC.

Start free