Anthropic released Opus 5 with fewer restrictions and Automatic Fallbacks
Anthropic introduced Opus 5 – according to the company, it is cheaper and less restricted than Fable 5, and even outperforms it on a number of benchmarks in its own announcement. The new Automatic Fallbacks beta feature redirects queries blocked by safety filters to a less capable model instead of returning an error message.
Anthropic released Opus 5 on Friday, the latest version of its flagship model. According to TechCrunch, it arrives just two months after Opus 4.8, available since 28 May, and after Mythos 5, Fable 5 and Sonnet 5 appeared in June. Haiku is still waiting for an upgrade to the 5 series.
According to the company, Opus 5 is “significantly better at verifying its own work and iterating carefully until it succeeds” – as an example from benchmark tests, Anthropic cites a situation where the model wrote its own computer vision pipeline based on an incomplete specification. Opus 5 is smaller than Fable 5, but according to the company, it is also cheaper and less restricted, and even outperforms Fable 5 on a number of benchmarks listed in the announcement.
Opus 5 has shed some of the restrictions that accompanied Fable 5 – like its predecessor Opus 4.8, it is not subject to the thirty-day data retention policy that applies to Fable and Mythos. Safety measures remain in place for cybersecurity tasks, such as exploit development or penetration testing: the model may look for vulnerabilities in source code (which the company considers a more defensive use), but may not scan for vulnerabilities in a software binary. According to the company, safety classification is expected to activate 85 % less often with Opus 5 than with Fable 5.
Anthropic is also launching the Automatic Fallbacks beta feature, which users can enable voluntarily. If a query triggers the safety classifier, the system automatically redirects it to a less capable model, so API users receive a usable response instead of an error message.
Why it matters
For developers working with the Anthropic API, this presents a concrete choice: according to the company, Opus 5 offers a lower price and fewer blocked requests than Fable 5, which may affect which model they choose for production deployment. The Automatic Fallbacks feature also reduces the risk that an application receives an error message instead of a result. For companies with data protection requirements, the difference in data retention policies between Opus 5 (without thirty-day retention) and Fable and Mythos is significant, and may play a role when choosing a model for sensitive environments.
Two audiences, two different impacts
What this means
For individuals
Developers using the Anthropic API now have access to a model that, according to the company, is cheaper and blocked less often than Fable 5, and they can enable Automatic Fallbacks so that queries caught by the safety filter do not result in an error.
For a business
Companies choosing between models from Anthropic must account for the different data retention policies (Opus 5 without thirty-day retention, unlike Fable and Mythos) and the lower frequency of safety blocks with Opus 5, which affects both costs and compliance decisions for sensitive deployments.
DevelopmentCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.