Anthropic Releases Opus 5.5: Fable 5.1-Level Performance and Lower Prices
Anthropic has launched Opus 5.5, offering performance on par with Fable 5.1, reduced token pricing, and improved usage limits.
Anthropic Releases Opus 5.5
Anthropic has announced the release of Opus 5.5, the latest iteration of its flagship AI model. Benchmark results indicate that Opus 5.5 delivers performance comparable to Fable 5.1.
Availability
The model is already accessible in:
- Chat interface
- API endpoints
Pricing
A significant update involves cost reduction. Compared to Opus 5, the pricing for Opus 5.5 has been lowered to:
- $4 per million input tokens
- $20 per million output tokens
Limits and New Features
Anthropic has also optimized usage constraints:
- Increased five-hour limits, allowing for more complex tasks without waiting periods.
- Introduction of «banked reset», a new mechanism that provides more flexibility in resetting or accumulating limits, enhancing resource efficiency for active users.
Why it matters
AnalysisThe combination of lower costs and enhanced limits makes Opus 5.5 a more viable option for developers and enterprises, while its competitive benchmark scores reinforce Anthropic's position in the high-performance LLM market.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Claude Fable 5: When to Use the Powerful Model and Which Effort Level to Choose
A comprehensive breakdown of Claude Fable 5 and the effort parameter: behavioral reasoning signals, level availability matrix (low to xhigh), per-turn cost calculations, and practical API/Claude Code setups.
Claude Opus (Flagship AI Model by Anthropic)
The most powerful heavyweight model in the Anthropic family, designed for complex philosophical and scientific analysis, multi-layered text generation, and deep contextual understanding.
Anthropic releases FlashProve in open source, a tool for provable inference
Anthropic published a prototype of an open cryptographic protocol called FlashProve that lets anyone verify a model response without access to model weights.
Hermes Agent: From Installation to Advanced Level
An autonomous AI assistant that lives on your server, learns from every conversation, works 24/7, and communicates via Telegram. Open source, MIT license, free.