Anthropic представила Opus 5: дешевле и менее ограниченная, чем модель Fable

Anthropic unveiled Opus 5 on July 24, 2026 — a new version of its flagship Claude Opus model line. According to TechCrunch, Opus 5 is simultaneously cheaper and less restricted in its responses than the Fable model, which, in the outlet's assessment, makes it the preferable choice in most use cases.

How Opus 5 differs from Fable

Opus 5 differs from Fable on two parameters named by TechCrunch: price and the degree of restrictions. Fable is a model from the same Claude generation that Anthropic directly compares the new release to. The outlet notes that Opus 5 will be cheaper, and its responses will lean less on safety filters — that is, the model refuses less often and holds firm boundaries only where it is genuinely needed.

• Announcement date — July 24, 2026
• Developer — Anthropic, Claude Opus line
• Opus 5 is cheaper than the Fable model
• Opus 5 is less restricted in its responses than Fable
• TechCrunch's conclusion — preferable in most use cases

The combination of "cheaper + fewer refusals" addresses two frequent complaints about top-tier models: high token costs and excessive caution. If both promises hold up in practice, Opus 5 becomes the default choice where teams previously had to choose between price and response quality.

Price is one of the main barriers to the mass adoption of powerful models: at large request volumes, token costs quickly become the primary expense item. Making the flagship cheaper directly affects where it makes sense for companies to use the top-tier model and where they have to switch to simpler models.

What "less restricted" means

"Less restricted," in TechCrunch's wording, means that Opus 5 less often blocks legitimate requests that previous models might have declined out of excessive caution. The essence is balance: fewer false refusals on harmless tasks while preserving boundaries on genuinely dangerous ones. For developers this is critical: an unnecessary refusal from the model breaks an automated workflow and requires manual intervention.

Anthropic has historically bet on a cautious tone for its models, and excessive caution has been a frequent complaint about the Claude line. The shift toward fewer restrictions in Opus 5 is a notable change in product policy, not just a technical version update.

Why there are so few details so far

In its announcement of July 24, 2026, Anthropic did not disclose specific token prices, benchmark results, or comparison tables with Fable. TechCrunch presents the news at the level of its core thesis: Opus 5 is cheaper and more permissive than Fable. The exact figures — cost, limits, test metrics — will apparently be provided by the company separately, based on the material.

"Opus 5 will be both cheaper and less restricted than

Fable, which will likely make it preferable in most use cases," per TechCrunch's wording.

For readers, this means Opus 5 should be judged by facts, not by the announcement. The key questions — exactly how much cheaper compared to Fable, on which tasks the restrictions have been lifted, and how this affects quality — remain open until the details are published and the first user tests are in.

What this means

Opus 5 is a signal that Anthropic is simultaneously pushing down on price and easing excessive restrictions, addressing the two main complaints about expensive flagship models. Teams that had to choose between cost and a model's "cooperativeness" now have reason to reconsider their default choice in favor of Opus 5 — as soon as figures and independent measurements appear.