Anthropic releases Claude Opus 5.5, faster and 40 per cent cheaper to run
Claude Opus 5.5 runs 30 per cent faster than Opus 5, costs less through the API and, Anthropic says, matches its restricted Fable model on many benchmarks.
Anthropic has released Claude Opus 5.5, the new top model in its Claude line-up. Launched on 22 September 2026, it is faster than Opus 5, cheaper to use and, according to Anthropic, matches or beats its more restricted Fable model across many benchmarks. For Australian businesses and developers building on Claude, it means more capability for less money, and for subscribers it arrives with higher usage limits.
What's new
Anthropic says Opus 5.5 generates output 30 per cent faster than Opus 5 and is about 40 per cent cheaper on typical workloads. TechCrunch reports the company describes it as "the strongest-performing model we've tested to date."
The company is pitching it at agentic coding, knowledge work, financial analysis and code review. Anthropic's published results include 66.4 per cent on Terminal-Bench 4.0, 54.4 per cent on FrontierCode, 57.8 per cent on CursorBench and 81.8 per cent on the OSWorld 2.0 computer use test. One early tester, Anthropic says, completed a 680,000-line code migration in under a day.
There is also a change in tone. TechCrunch notes the model uses less jargon and puts the most important information at the start of its replies, which should make longer answers easier to scan.
Safety settings
Opus 5.5 carries the same cybersecurity and biology safeguards as Fable 5.1, limiting its use for tasks such as finding software exploits. Anthropic runs verified access programs for cybersecurity professionals and life sciences researchers who need to go further. The company also reports an 85 per cent reduction in attempts to circumvent containment boundaries compared with Opus 5.
The release is Anthropic's first since chief executive Dario Amodei called for the industry to pace capability gains so safety research can keep up, a context TechCrunch highlights in its coverage.
What else is coming
Opus is the largest of Anthropic's three model tiers. TechCrunch reports Sonnet 5.5 and Haiku 5.5 will follow "in the coming weeks" with similar improvements. OpenAI, meanwhile, launched its GPT-6 Sol and Luna models roughly 90 minutes after Anthropic's announcement.
Price and availability
- Where: Claude apps and the Claude Platform, plus Amazon Web Services, Google Cloud and Microsoft Azure. The API model name is claude-opus-5-5.
- API pricing: US$4 per million input tokens and US$20 per million output tokens (both 20 per cent less than Opus 5). Cache reads cost US$0.20 per million tokens and cache writes US$5.
- Subscriptions: Anthropic is increasing five-hour usage limits on Pro, Max, Team and seat-based Enterprise plans, and giving subscribers a rate limit reset.
- Australia: no local pricing has been announced; API usage is billed in US dollars.
Sources: Anthropic and TechCrunch
