Anthropic releases Opus 5.5 with lower prices and Fable-level performance
First reported by TechCrunch ·
The cost of accessing top-tier AI models has decreased, making advanced capabilities more affordable for users and developers.
Anthropic has released Opus 5.5, its latest AI model, demonstrating a new state-of-the-art performance in coding and knowledge work. The new model reportedly outperforms its larger Fable model on several benchmarks and informal tasks. Opus 5.5 is also priced lower than its predecessor, with output tokens costing $20 per million tokens compared to $25 for Opus 5. The model is also faster and has reduced compute requirements. Anthropic plans to release Sonnet 5.5 and Haiku 5.5 with similar improvements in the coming weeks. This release follows Anthropic CEO Dario Amodei's recent advocacy for pacing AI capabilities advancement alongside alignment efforts. Opus 5.5 is subject to the same safety restrictions as the Fable model, limiting its use in areas like exploit discovery and biological weapon development. Anthropic stated it is investing in infrastructure for more advanced safety training and evaluation systems for future models.
Anthropic's release of Opus 5.5 signifies a competitive shift in the AI model market, with significant performance gains coupled with reduced pricing. This move challenges competitors to match both capability and cost-effectiveness, potentially driving down prices across the board and accelerating adoption of more advanced AI tools for a wider range of applications. The emphasis on improved communication and reduced jargon also suggests a move towards more user-friendly AI interfaces.
The company's simultaneous focus on safety and responsible AI development, as highlighted by CEO Dario Amodei's stance on pacing capabilities with alignment, indicates a growing industry awareness of the need for robust safeguards. As Anthropic invests in advanced training and evaluation systems, this sets a precedent for how future powerful AI models will be developed and deployed, with public policy likely playing an increased role in ensuring system safety.
AI-written summary. May contain errors.