Claude Opus 5.5 Is Now Available in Cosmic

Claude Opus 5.5 is available in Cosmic. Select it from the model picker in the dashboard, or pass claude-opus-5-5 as the model in the AI text generation API.

Anthropic released Opus 5.5 on September 22, 2026. It is their new flagship, and unusually for a flagship release, it costs less than the model it replaces.

What it costs

Pricing read from Anthropic's pricing page on September 22, 2026:

• Input: $4 per million tokens. Opus 5 was $5.
• Output: $20 per million tokens. Opus 5 was $25.
• Cache reads: $0.20 per million tokens. Opus 5 was $0.50, so reading cached context costs 60% less.
• Cache writes: $5 per million tokens. Opus 5 was $6.25.

On the same page, Anthropic describes Opus 5.5 as the "daily driver for agentic coding and enterprise work" and has moved Opus 5 down into the legacy models section. Fast mode is available for Opus 5.5 at up to 2.5x speed for 2x standard pricing.

If you want the benchmark numbers and the cost math on a realistic monthly workload, we covered those separately in Claude Opus 5.5 vs Opus 5.

Using Opus 5.5 in Cosmic

Text generation runs through the same REST API and TypeScript SDK you already use for content. Pass the model string and Cosmic handles the provider call:

import { createBucketClient } from '@cosmicjs/sdk';

const cosmic = createBucketClient({
  bucketSlug: process.env.COSMIC_BUCKET_SLUG,
  readKey: process.env.COSMIC_READ_KEY,
  writeKey: process.env.COSMIC_WRITE_KEY,
});

const response = await cosmic.ai.generateText({
  prompt: 'Write a product description for a wireless keyboard.',
  model: 'claude-opus-5-5',
});

AI generation draws tokens from your plan's allocation, and different models draw at different rates. See the AI API reference for the current model list and token accounting.

Why the model string belongs in your content layer

Anthropic has shipped two frontier models inside a month and has announced Sonnet 5.5 and Haiku 5.5 for the coming weeks. Every one of those releases changes the price and capability tradeoff you made last time.

If the model your agents use is compiled into application code, each release is a pull request, a review, and a deploy. If it is a field on a content object, it is an edit. Cosmic gives you a structured, versioned content store that your agents read and write through a REST API, a TypeScript SDK, and an MCP server, so swapping models is a content change rather than a release.