Anthropic Launches Claude Opus 5.5 With 40% Lower Costs Ahead of $2T IPO

Anthropic has launched Claude Opus 5.5 as it cuts model costs and prepares for a November public listing. The release combines lower pricing, faster output, and stronger coding results while competition intensifies among leading AI developers.

Claude Opus 5.5 Targets Lower Costs and Faster Enterprise Work

Claude Opus 5.5 is Anthropic’s new family of Claude 5.5, optimized for coding, research, and professional knowledge work. According to Anthropic, the model’s performance is close to Claude Fable 5.1 with reduced spending for common tasks.

The company estimates that Opus 5.5 costs about 40% less to run than Opus 5. The input pricing is $4 per million tokens, and the output pricing is $20 per million tokens.

Cached reads now cost $0.20 per million tokens, representing a 60% reduction from Opus 5. According to Anthropic, output generation has also accelerated by over 30%.

Those changes come in the wake of AI customers demanding improved models, but not at the rapid expense of inference costs. OpenAI has followed the same market pressure with cheaper additions to its GPT-6 family.

On Tuesday, just like Anthropic, OpenAI introduced GPT-6 Sol and GPT-6 Luna with API pricing below previous promotional GPT-5.6 levels. As a result, both firms are dealing with cheaper competition from Alibaba, Moonshot AI, and DeepSeek.

Coding Gains Arrive During Wider AI Safety Debate

In addition to cost savings, there are improvements in several coding and professional benchmarks, such as Terminal-Bench 4.0, where Claude Opus 5.5 achieved a 66.4% score, while Opus 5. only netted 52.3%.

The model also achieved a score of 57.8% on CursorBench 4.0 and 1,846 Elo on GDPval-AA. Anthropic uses these tests to measure software development and professional knowledge performance.

Long-running coding work remains another focus for the release, with one tester completing a 680,000-line code migration in less than one day. Another tester audited and repaired a 200,000-line codebase in less than three hours. Anthropic reported it took over 20 hours to process that amount of work with Opus 5.

The launch also coincides with a broader discussion on the pace of deployment of frontier AI systems. Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and Elon Musk have urged more safeguards and a slower development tempo.

Despite this, US President Donald Trump has taken a different approach, opposing new rules that could slow U.S. AI development. At the UN, Trump said he would not “stifle growth” in technology he expects to exceed the Industrial Revolution but the Justice Department would keep an eye on the sector and take action if needed.

Anthropic IPO Plans Add Business Focus to Opus 5.5 Launch

The commercial push follows news that Anthropic is reportedly working on a potential November IPO. As we reported, the Claude developer could seek a valuation near $2 trillion.

The company could also raise to $100 billion, though those figures remain subject to change. However, so far, Anthropic has not publicly confirmed final IPO terms or a completed valuation.

Lower model costs could matter as Anthropic expands enterprise use before any public-market debut. The company is also expanding its distribution channel via cloud providers and its own Claude platform.

As of now, Claude Opus 5.5 is available through Claude, Amazon Web Services, Google Cloud, and Microsoft Azure. In the meantime, Anthropic plans to release Claude Sonnet 5.5 and Claude Haiku 5.5 in the coming weeks.