Claude Opus 4.5 Released: Pricing, Availability, and What’s New (Dec 2025)

What Anthropic Shipped

• Model: claude-opus-4-5-20251101 (200K context, tool use, images, text).
• Pricing (API): $5 per 1M input tokens; $25 per 1M output tokens; prompt caching and batch discounts apply (per claude.com/pricing).
• Availability: Claude web/desktop apps (Pro/Max/Team/Enterprise), Claude API, Amazon Bedrock, Google Vertex AI, and AWS Foundry.
• Use cases: Difficult multi-file refactors, deep debugging, long-context analysis, and high-stakes reasoning where quality matters more than cost.

Opus 4.5 vs Sonnet 4.5

• Opus 4.5: Highest quality, 200K context, best for complex refactors/architecture/debugging; higher cost.
• Sonnet 4.5: Lower cost, also 200K context, better for day-to-day coding, chat, and iterative work.

Features and Behavior

• Full tool calling and JSON output support.
• Vision: accepts images alongside text.
• Prompt caching and batch routes reduce costs for repeated/system prompts.
• No computer-use/desktop control advertised for Opus 4.5 (standard tool use only).

How to Access

• Claude apps: Select Opus in Claude Pro/Max/Team/Enterprise.
• API: Use model ID claude-opus-4-5-20251101 with the Claude API; caching/batch optional.
• Partners: Bedrock/Vertex/Foundry catalogs list Opus 4.5 where enabled; billing via partner rates.

When to Choose Opus 4.5

• Multi-file or architecture-level changes where correctness is critical.
• Long-context reviews (design docs, large repos) within 200K tokens.
• Investigations that benefit from stronger reasoning (tradeoff: higher output cost).

Sources (December 2025)

• Claude pricing/models: https://claude.com/pricing
• Claude API docs: https://docs.claude.com/
• Partner availability (Bedrock/Vertex/Foundry): provider catalogs

Need another comparison? Check our directory for Claude, OpenAI, Google, DeepSeek, and local model options.