Skip to content
API Reference

Claude

Claude models offer long context windows, strong reasoning, and extended thinking. They support tools, structured output, and prompt caching. Call them through any text format — including the native Messages API.

Models

Model Code Notes
Claude Opus 5.5 claude-opus-5.5 Latest flagship; 1M context, adaptive thinking.
Claude Fable 5.1 claude-fable-5.1 Extends Claude Fable 5 at the same prices; stronger long-running agentic coding and research.
Claude Fable 5 claude-fable-5 Most capable widely released model; always-on thinking.
Claude Opus 5 claude-opus-5 Flagship; adaptive thinking on by default.
Claude Sonnet 5.5 claude-sonnet-5.5 Balanced; adaptive thinking.
Claude Sonnet 5 claude-sonnet-5 Balanced; adaptive thinking.
Claude Opus 4.8 claude-opus-4.8 Adaptive thinking.
Claude Opus 4.7 claude-opus-4.7 Adaptive thinking.
Claude Opus 4.6 claude-opus-4.6 Adaptive thinking.
Claude Opus 4.5 claude-opus-4.5 Non-adaptive thinking (budget_tokens).
Claude Sonnet 4.6 claude-sonnet-4.6 Balanced; adaptive thinking.
Claude Sonnet 4.5 claude-sonnet-4.5 Balanced; non-adaptive thinking.
Claude Haiku 4.5 claude-haiku-4.5 Fastest; non-adaptive thinking.

Capabilities

  • tools, tool_choice, parallel_tool_calls
  • temperature, top_p, max_tokens, response_format — send only one of temperature and top_p; Claude 4.7 and later ignore temperature (other than 1), top_p and top_k
  • Extended thinking via thinking or reasoning_effort — normalised per model.
  • Prompt caching with cache_control — see Prompt Caching.
  • Assistant prefill is not supported on Claude 4.6 and later.

See Per-Model Details → Claude 4+ for thinking modes and budget_tokens rules.

Example

from openai import OpenAI

client = OpenAI(base_url="https://llm.siraya.ai/v1", api_key="<API_KEY>")
resp = client.chat.completions.create(
    model="claude-opus-4.8",
    messages=[{"role": "user", "content": "Summarise this contract."}],
)
print(resp.choices[0].message.content)