How to Use Claude Opus 4.5 (Dec 2025)

Claude Opus 4.5 is Anthropic's previous-generation premium model with a 200K token context window, $5/$25 MTok pricing, tool use, and vision capabilities. It is available via the Claude app, API (model ID claude-opus-4-5-20251101), and cloud partners (Bedrock, Vertex AI, Foundry). Anthropic has since released Opus 4.6 with 1M context and improved quality at the same price.

TL;DR

• Access Opus 4.5 via the Claude app (Pro/Max/Team/Enterprise), API with model ID claude-opus-4-5-20251101, or cloud partners (Bedrock, Vertex AI, Foundry).
• Pricing is $5/MTok input and $25/MTok output via the Claude API; partner clouds use their own rate cards.
• If you have access to Opus 4.6, use it instead -- it offers improved quality and a 1M token context window at the same price.
• Opus 4.5 supports tool use (function calling) and vision (image + text inputs).
• Use prompt caching and batch routes for cost optimization on repeated prompts.

Note: This guide covers the December 2025 Opus 4.5 release. Anthropic has since released Claude Opus 4.6 with improved quality and a 1M token context window. For current model guidance, see our Claude vs GPT vs Gemini comparison and the Claude API profile.

Quick Reference

Using Opus 4.5 in the Claude App

1. Sign in at claude.ai with a Pro, Max, Team, or Enterprise account.
2. Start a new chat and select Opus from the model picker.
3. Paste your task or attach files/images. Keep total input under ~200K tokens.
4. For coding tasks, include the relevant files and be explicit about what you want changed.

Calling Opus 4.5 via the API

Endpoint: https://api.anthropic.com/v1/messages

curl https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-5-20251101",
    "max_tokens": 800,
    "messages": [
      {
        "role": "user",
        "content": "Review this function for edge cases:\n```python\n...code...\n```"
      }
    ]
  }'

Tips for API usage:

• Keep total tokens (prompt + response) within 200K
• Use the system parameter to set style and constraints---keep it stable to benefit from prompt caching
• For structured actions, include a tools array and let Claude return tool_use calls
• Enable prompt caching or batch routes for repeated prompts to reduce cost

Using Opus 4.5 on Cloud Partners

Partner pricing differs from the direct Claude API. Check each provider's rate card.

Prompting Best Practices

These apply to all Claude models, not just Opus 4.5:

1. Be explicit about goals, constraints, and the files or APIs involved.
2. Use structure: Bullet lists and numbered steps reduce ambiguity.
3. Trim context: Include only the relevant file sections. Stay within the context window.
4. Plan first: For multi-step work, ask for a short plan before executing.
5. Use tool calls when you need structured actions instead of free-text responses.
6. Cache system prompts to reduce cost and latency on repeated calls.

Opus 4.5 vs Sonnet 4.5

Sources (December 2025)

• Claude pricing and models: anthropic.com/pricing
• Claude API reference: docs.anthropic.com
• Claude tool use docs: docs.anthropic.com/docs/en/build-with-claude

For current model guidance, see our beginner's guide to AI models and directory.