Anthropic launches Opus 5.5 at reduced prices, delivering Fable-tier performance
Anthropic launched Opus 5.5 just two months after Opus 5, claiming state-of-the-art coding performance that beats the larger Fable model. The new release is cheaper ($20 per million output tokens vs. $25) and faster, with clearer communication style. Safety measures matching Mythos-level capabilities remain in place. The launch follows CEO Dario Amodei's call for pacing frontier AI progress, with Sonnet 5.5 and Haiku 5.5 expected soon.
Anthropic released Opus 5.5 on Tuesday, and the company says the new model sets a fresh state of the art in coding and knowledge work. According to Anthropic, the release even beats the larger Fable model on many benchmarks and completed several informal tasks that Fable could not finish. The launch arrives only two months after Opus 5 debuted on July 24th, signaling an unusually fast release cadence for the company's flagship line.
Lower Prices, Faster Performance
Opus 5.5 is meaningfully cheaper than the model it replaces. Output tokens are priced at $20 per million tokens, down from $25 for the previous version, and other pricing tiers show similar reductions. The model also runs faster, a change Anthropic attributes to a lower overall compute requirement for serving it.
Beyond cost and speed, Anthropic has reworked how the model communicates. Opus 5.5 is less inclined to fall back on jargon and more likely to place key information at the beginning of its responses, making its output easier to scan.
Safety Safeguards Stay in Place
Anthropic rates Opus 5.5 as comparable to Mythos in biology and cybersecurity capabilities, which means it falls under the same protective measures as the Fable model. Those restrictions cap how far the models can go in areas such as finding exploits in compiled programs or developing recognizable biological weapons.
The safety training process resembled that of earlier models: alignment testing plus pre-release evaluations by external groups including METR and Frontier Design. Still, Anthropic stressed that more advanced training and evaluation systems are already in the works for future models, along with stronger security and monitoring infrastructure.
"A capacity takes time to build, and we've started to put the infrastructure in place to support it," the company wrote in its blog post. "We expect to share more details on these efforts soon."
A New Era of Pacing
This is Anthropic's first model launch since CEO Dario Amodei publicly endorsed the idea of pacing the frontier — deliberately slowing capability progress so that safety work can keep pace. "I have become convinced that fully addressing the risks requires even more prudence," Amodei wrote earlier this month, "not just investing in risk prevention, but pacing the rate of capabilities advancement so that risk prevention has time to keep up."
Attention now turns to Sonnet 5.5 and Haiku 5.5, which Anthropic says will arrive "in the coming weeks" with similar performance gains.
Comments
No comments yet. Be the first to comment.