Anthropic releases Opus 5.5 with lower prices and Fable-level performance
Anthropic released Opus 5.5 on September 22, 2026, cutting output token prices to $20 per million and delivering coding benchmarks that surpass the larger Fable model.
Stock photo for illustration only, not from the actual event
- Anthropic launched its newest AI model, Opus 5.5, on September 22, 2026.
- Output token pricing dropped to $20 per million tokens from the previous $25.
- Outpaces the larger Fable model across multiple standard benchmarks and informal tasks.
- Sonnet 5.5 and Haiku 5.5 are scheduled to be released in the coming weeks.
Anthropic released its newest artificial intelligence model, Opus 5.5, on Tuesday, September 22, 2026, establishing a state-of-the-art benchmark for coding and knowledge work performance. The company reported that the new model runs faster and requires less underlying compute resources compared to its predecessor version.
In performance evaluations, Opus 5.5 outpaces the larger Fable model in numerous benchmark tests and successfully handles a series of informal tasks that Fable failed to complete. Furthermore, the model incorporates communication updates, reducing its reliance on corporate or technical jargon while prioritizing critical information at the beginning of its generated messages.
Along with performance upgrades, the new version is significantly more affordable than its predecessor. Output tokens for Opus 5.5 are priced at $20 per million tokens, down from $25 for the previous model, with similar price reductions applied to other metrics. This launch arrives just two months after the July 24 release of Opus 5, with Sonnet 5.5 and Haiku 5.5 slated to debut in the coming weeks.
Stock photo for illustration only, not from the actual event
Anthropic's ability to lower operational costs while boosting performance in version 5.5 highlights significant progress in model optimization, reducing the hardware footprint required to serve large language models. This trend mirrors broader industry shifts where AI providers increasingly compete on cost-efficiency and economic viability, enabling enterprises to deploy advanced models at scale.
Regarding safety measures, Anthropic noted that Opus 5.5 matches the biology and cybersecurity capabilities of Mythos, subjecting its release to the same rigorous safeguards as the Fable model. These restrictions limit how the system can be utilized for discovering software vulnerabilities or developing biological threats. Safety training involved pre-release evaluations by independent organizations including METR and Frontier Design.
Source: TechCrunch
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment