Opus 5.5 Price: $4 Input, $20 Output per Million Tokens

Pricing checked on September 26, 2026 against the public BazaarLink model catalogue and Anthropic’s model documentation.

Price data as of 2026-10-09 (checked against the BazaarLink model list; the rates match the table below). Actual BazaarLink billing follows the model page and your invoice.

Claude Opus 5.5 costs $4 per million input tokens and $20 per million output tokens, with a 1,000,000-token context window. The list price is 20% below Opus 5 ($5 and $25). Prices are in US dollars. See the live price and parameters on the Opus 5.5 model page, and other models at /en/models.

Claude Opus 5.5 API price table

The catalogue uses Claude Opus 5.5 as the model ID. The price is token based: each request combines the input-token charge and output-token charge. Cache rates apply only to the matching cache operation; they are not a replacement for the normal input rate.

USD cost examples

Use this estimate:

total = input_tokens / 1,000,000 × 4 + output_tokens / 1,000,000 × 20

These examples exclude cache operations. A stable system prompt or other repeated prefix can be evaluated for caching separately, using the published write and read rates.

Context and maximum output

Anthropic’s current model page lists a 1M-token context window and a 128K-token maximum output. Context is the amount the request can make available to the model; it is not a flat charge. Your actual bill still depends on the input and output tokens used.

Calling it with an OpenAI-compatible SDK

The request shape below uses the OpenAI Python client with a BazaarLink-compatible base URL. Keep the API key in an environment variable.

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.bazaarlink.ai/v1",
    api_key=os.environ["BAZAARLINK_API_KEY"],
)

response = client.chat.completions.create(
    model="claude-opus-5.5",
    messages=[
        {"role": "user", "content": "Summarize this document in five bullets."}
    ],
)

print(response.choices[0].message.content)

Keep max output large enough for the answer you actually need. If you are sending long repeated context, compare the cache-write and cache-read charges with the normal input charge before changing the request shape.

What one task and one month cost

All figures are arithmetic on assumed workloads, not measurements, and ignore caching discounts and retries.

One request: 50,000 input tokens and 5,000 output tokens.

One month: 1,000 requests a day, 10,000 input and 1,000 output tokens each, 30 days, so 300M input and 30M output tokens. Opus 5.5 comes to 300 x $4 + 30 x $20 = $1,800 (Opus 5: $2,250); Sonnet 5.5 is $900; Haiku 5.5 is $45.

The Haiku 5.5 figures are the BazaarLink model list rates; per Anthropic's announcement, Haiku 5.5 prompts above 100,000 tokens are $0.50 input and $2.50 output. Actual BazaarLink billing follows the model page and your invoice. Reasoning effort changes output volume: thinking tokens are billed as output, so run a small pilot before you set a budget.

Is Opus 5.5 worth it?

Opus 5.5 is easiest to justify when the task benefits from long context, sustained reasoning, or careful coding and knowledge work. For a direct capability comparison, see Claude Opus 5.5 vs Opus 5. If the task is short, repetitive, and cost-sensitive, calculate the token mix first; a lower-priced model may be a better fit.

Sources

• BazaarLink public model catalogue
• Anthropic Claude Opus 5.5 model overview

TWD billing · Taiwan invoices · leading AI models · OpenAI-compatible API