Anthropic official brand assets (anthropic.com)
Claude Opus 5.5 undercuts its predecessor by 20% as the AI safety company eyes a public listing that could value it above $1 trillion
Anthropic just dropped the price of frontier intelligence. The AI safety company released Claude Opus 5.5 on September 22, pricing it at $4 per million input tokens and $20 per million output tokens, a 20% discount compared to Opus 5, which launched barely two months earlier on July 24.
The move isn’t charity. It’s strategy. Anthropic is preparing for an IPO that could push its valuation past $1 trillion, and nothing says “growth story” to prospective public market investors quite like a model that’s both better and cheaper than the last one.
What Opus 5.5 actually delivers #
The new model reportedly rivals or exceeds the performance of Anthropic’s larger Fable models on coding and knowledge work benchmarks, despite requiring less compute to run.
Beyond raw performance, Anthropic highlighted improvements in communication style, speed, and safety alignment. The company says Opus 5.5 is 85% less likely to attempt bypassing safety boundaries compared to earlier models.
Opus 5, which launched in late July, was already positioned as offering near-Fable 5 performance at roughly half the price. Now Opus 5.5 pushes that value proposition further, shaving another fifth off the cost while boosting capability.
AI, tech, and the markets they move—in one daily briefing.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
The IPO math #
Anthropic’s path to public markets has been building momentum for months. The company raised a staggering $65 billion in May 2026, a round that valued it at $965 billion post-money. A confidential S-1 filing reportedly followed around June. Internal discussions have centered on IPO valuations that could exceed $1 trillion.
A market-wide race to the bottom on pricing #
Anthropic isn’t cutting prices in a vacuum. The entire AI model market has been experiencing sustained pressure from customers gravitating toward cheaper alternatives. Open-weight models and low-cost providers, particularly from Asian competitors, have been steadily eroding the pricing power that frontier labs once enjoyed.
The company isn’t done, either. Additional models in the lineup, including Sonnet 5.5 and Haiku 5.5, are expected to arrive soon. Those models typically serve different price-performance tiers, with Haiku historically targeting the cheapest, fastest use cases and Sonnet occupying the middle ground.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our