claude-opus-5 API
Claude Opus 5 is an Anthropic Opus version for software engineering, analysis and multi-step execution, emphasizing verification and revision. It is now an active legacy model. Use XiuAI to compare current rates, protocols and migration requirements for Opus 5.5 against your tasks.[1][2][4]
claude-opus-5Model overview
Published provider specifications. Available features depend on the current API route.
- API routes
- OpenAI-compatible API / openai-response / openai-response-compact / Anthropic Messages / Gemini GenerateContent / openai-alpha-search
Engineering execution and verification
Consider workflows that ask the model to analyze a problem, make a change and inspect the result: root-cause investigation, code editing, research synthesis or reviewable documents. Explicit completion criteria and tool boundaries make outputs easier to verify than a general request to improve them.[2]
The release shows visual, scientific and business tasks completed with code and tools. Those examples illustrate possible uses; they do not mean the model supplies the execution environment or guarantee success on every task.[2][1]
Key differences when moving to Opus 5.5
Opus 5 accepts requests with thinking disabled; Opus 5.5 always uses adaptive thinking. Review thinking settings, tool choice and response parsing during migration, and budget output for both thinking and text. Comparing list rates for the same input length is not enough.[3]
API access
Calling through XiuAI and comparing costs
Use claude-opus-5, your existing balance and an authorized key with a protocol shown on this page. Verify responses and tool loops using representative tasks before deciding to retain this version or switch.[4][3]
Effort affects processing depth and usage, with results depending on the task. Evaluate total cost using current input, output and cache rates plus request records. Historical vendor price comparisons or benchmark efficiency claims are not XiuAI guarantees.[2][1][4]
Send a request using your existing API key.
POST /v1/chat/completionscurl 'https://router-api.xiu.ai/v1/chat/completions' \
-H "Authorization: Bearer $XIUAI_API_KEY" \
-H 'Content-Type: application/json' \
--data-raw '{
"model": "claude-opus-5",
"messages": [
{
"role": "user",
"content": "Hello"
}
]
}'Read the API documentationFrequently asked questions
Can Opus 5 and Opus 5.5 use identical thinking settings?
Not without checking. The migration guide states that 5 accepts disabled thinking, whereas 5.5 rejects disabled and manual-budget modes. Review effort and output budgets for the target model, and parse content blocks by type.[3]
Related models
claude-opus-5-5claude-sonnet-5-5Sources and updates
Specifications reference provider documentation; prices and availability come from the current XiuAI catalog.
Content reviewed: 2026-11-16
[1] Claude Opus 5 model reference[2] Claude Opus 5 announcement[3] Anthropic migration guide[4] Current XiuAI model and pricing dataModel documentationTool callingStructured outputContext documentation