Anthropic released Claude Opus 5.5 on Tuesday, Sept. 22, the first model in its new Claude 5.5 family, and cut Opus API prices by 20%. In its announcement, the company said the model performs at the level of Claude Fable 5.1 on most work and, at default settings, costs 40% less to run than Opus 5 on typical workloads. It is Anthropic's first release since CEO Dario Amodei called for the industry to slow the pace of AI development 10 days earlier.
Pricing and speed
Opus 5.5 costs $4 per million input tokens and $20 per million output tokens, down from $5 and $25 for Opus 5. Cache reads fall 60%, to $0.20 per million tokens from $0.50, and cache writes fall to $5 from $6.25. Anthropic said the model generates output more than 30% faster than Opus 5. A fast mode in Claude Code and the Claude Platform runs at up to 2.5 times the speed and costs $8 per million input tokens and $40 per million output tokens.
Developer Simon Willison noted that Opus 4.5 through Opus 5 all shared the $5/$25 price. According to Artificial Analysis, the model keeps the 1 million token context window of Opus 5 and supports image and text input.
Benchmarks
Anthropic's published results put Opus 5.5 at 66.4% on Terminal-Bench 4.0, against 55.8% for Fable 5.1, 52.3% for Opus 5 and 57.9% for OpenAI's GPT-6 Astra. On GDPval-AA v2.1, a test of work across 44 occupations, it scored 1,846 Elo, compared with 1,735 for Fable 5.1 and 1,708 for Opus 5. It scored 81.8% (partial credit) on OSWorld 2.0 for computer use and 67.7% on Humanity's Last Exam with tools. On Zapier's AutomationBench, Opus 5.5 scored 40.0% to GPT-6 Astra's 41.4%. Anthropic said benchmark margins have become "a less reliable guide to real-world differences" at this level of capability.
Independent tester Artificial Analysis scored Opus 5.5 at 58 on its Intelligence Index at max effort, the top score on the index. On its own Terminal-Bench 4.0 run, Artificial Analysis measured Opus 5.5 at 59.6%, level with GPT-6 Astra and 11 points above Opus 5. The firm also found that at max effort Opus 5.5 used about 1.6 times as many output tokens per task as Opus 5, leaving its cost per task level with its predecessor on that index.
Availability and safeguards
The model is available on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure, under the model ID claude-opus-5-5. Anthropic raised five-hour usage limits on Pro, Max, Team and seat-based Enterprise plans and gave subscribers a rate limit reset, which MacRumors reports can be used through Oct. 22. Claude Sonnet 5.5 and Claude Haiku 5.5 will follow in the coming weeks, the company said.
Anthropic said Opus 5.5 is comparable to Claude Mythos 5.1 in biology and cybersecurity, so it ships with safeguards similar to those on Fable 5.1; most cybersecurity tasks are re-routed to Opus 4.8. Vetted organizations can apply to a new Life Sciences Verification Program for biology research access, and the company plans to expand its Cyber Verification Program to include Opus 5.5. The model also launches with "preserved thinking," an anti-distillation measure that stops API users from editing Claude's prior context to extract its reasoning; it applies to API accounts created on or after Aug. 31, 2026. Opus 5.5 is no longer available with thinking mode switched off. External evaluators including METR tested it before release.
The pacing debate
"We must slow the pace at which we improve the capabilities of AI models," Amodei wrote in "We Must Pace the Frontier," an essay The Register dates to Sept. 12. In the essay, he wrote that pacing "does not mean halting model training or technical progress," and committed Anthropic to giving embedded third-party evaluators employee-like access. The Register notes Opus 5.5 arrived 21 days after Fable 5.1 and Mythos 5.1, and that both Anthropic and OpenAI are preparing for initial public offerings, with Anthropic's possible before the end of the year and OpenAI's not expected until 2027.
CNBC reports the debate began when former Anthropic researcher Jacob Coxon posted to X on Sept. 8 that he had quit, and that OpenAI CEO Sam Altman and Tesla and SpaceX CEO Elon Musk joined Amodei's call. CNBC also reports both labs face pressure from cheaper open-weight models from Chinese firms including Alibaba, Moonshot AI and DeepSeek.
OpenAI's same-day release
OpenAI announced GPT-6 Sol and GPT-6 Luna roughly an hour after Anthropic's release, according to 9to5Google. Both sit below GPT-6 Astra, which debuted Sept. 3. Artificial Analysis reports Sol's price drops to $2/$10 per million input/output tokens from $4/$20 for GPT-5.6 Sol, and Luna's to $0.10/$0.50 from $0.20/$1.20. GPT-6 Astra and Claude Fable 5.1 both remain priced at $10/$50, per Willison.
Opus 5.5 follows the launch of Claude Opus 5 in July. Uristocrat previously reported that Anthropic expects its IPO to match or exceed SpaceX's $75 billion record raise, and that the company built an in-house silicon team to design chips for Claude.
Source: Anthropic, Artificial Analysis, The Register, CNBC
Comments