Claude Opus 5.5
AnthropicFrontier
The most capable Claude, for the hardest work.
- Pro
- Ultra
- Context window
- 1M tokens
- Max output
- 128K tokens
- Input, per 1K tokens
- 200 credits
- Output, per 1K tokens
- 1,000 credits
About Claude Opus 5.5
Anthropic's frontier model for long, complex tasks. Opus holds a plan over many steps, reasons carefully before it acts and produces the most reliable results when an agent has real work to finish on its own.
Best for
- Long autonomous tasks
- Complex analysis and planning
- Work where mistakes are costly
Capabilities
- YesTool useWorks with the connectors you assign to its agent.
- YesStructured outputAnswers in an exact data format when a task needs one.
- YesVisionReads images such as screenshots and photos.
- YesPrompt cachingReads repeated context from the cache at a lower rate.
Effort
Effort sets how much Claude Opus 5.5 reasons before it answers. Higher levels give it more room to think, which can use more credits.
- Low
- Medium (default)
- High
- Extra high
- Max
Each agent can set its own level; new agents start at medium.
What it costs. In credits, from your one balance.
A typical message, 8,000 tokens in, 6,000 of them read from the cache, and 800 tokens out, comes to about 1,260 credits on Claude Opus 5.5. Real messages vary with the length of the prompt and the reply.
Credits per 1K tokens
- Input
- 200
- Output
- 1,000
- Cache read
- 10
- Cache write
- 250
Typical messages a month
- Pro, 250,000 credits
- ≈ 198
- Ultra, 1,500,000 credits
- ≈ 1,190
Other models. Every agent can run a different one.
Questions about Claude Opus 5.5
How many credits does Claude Opus 5.5 use?
200 credits per 1K input tokens, 1,000 per 1K output tokens and 10 per 1K tokens read from the cache. A typical message (8,000 tokens in, 6,000 of them read from the cache, and 800 tokens out) comes to about 1,260 credits.
Which plans include Claude Opus 5.5?
Pro and Ultra. On Pro, the 250,000 monthly credits cover about 198 typical messages. On Ultra, the 1,500,000 monthly credits cover about 1,190 typical messages.
How much can Claude Opus 5.5 read and write?
It reads up to 1M tokens of context in a request and writes up to 128K tokens in one reply.
Can Claude Opus 5.5 work with images?
Yes. It takes images as input alongside text, so an agent can read the screenshots and photos you send it.
What does the effort setting do?
Effort sets how much Claude Opus 5.5 reasons before it answers. It offers low, medium, high, extra high, and max, and agents start at medium. Higher effort gives it more room to think, which can mean more output tokens and so more credits.