anthropic has introduced Claudius Sonnet 5a new model that offers capabilities previously exclusive to the Opus series at a more affordable price.
The launch occurs as The US Department of Commerce lifts export restrictions on Claude Fable 5 and Mythos 5. Starting Wednesday, Anthropic will restore global access to these two models.
Sonnet 5 is available with an introductory API price of $2 per million input tokens and $10 per million output tokens until August 31, 2026.
After that, the standard price will be $3 per million input tokens and $15 per million output tokens. The version is now available for Free, Pro and Max subscribers.
What makes Claude’s Sonnet 5 different?
Anthropic describes Sonnet 5 as the most agentive Sonnet model yet. The model introduces capabilities that were previously mostly limited to Opus 4.8, such as multi-step planning, use of tools including browsers and terminals, self-checking its output without explicit prompts, better instruction following, and improved agent search across large codebases.
These improvements are aimed at users who rely on Claude to code, research, automate, and work with documents.
Overall, Sonnet 4.6 was considered weaker at planning and understanding large codebases, but Sonnet 5 appears to close that gap significantly.
The pricing model positions the Sonnet 5 as a budget alternative to the Opus 4.8 when performance is considered. During the introductory period until August 31, 2026, Sonnet 5 costs $2 per million tokens in and $10 out.
The standard version is priced at $3 per entry and $15 per exit per million tokens. In comparison, Opus 4.8 costs $5 per million tokens in and $25 out. At standard price,
Sonnet 5 is about 60% cheaper than Opus 4.8. Users can choose between the two depending on their preferences for cost savings or maximum performance.
Benchmark tests from BrowseComp and OSWorld-Verified back up the performance claims, indicating that Sonnet 5 performs noticeably closer to Opus 4.8 than Sonnet 4.6.
Why Fable 5 and Mythos 5 are returning
The release of Sonnet 5 comes as Anthropic restores access to Claude Fable 5 and Mythos 5.
These models had been offline for two and a half weeks following a US government order restricting access by foreign citizens, citing national security concerns.
Anthropic announced in X that the Department of Commerce lifted export controls on both models. Starting Wednesday, access will be restored to global users and business customers.
Commerce Secretary Howard Lutnick said the government had worked closely with Anthropic over the past two weeks to evaluate and approve Fable 5 and support American leadership in AI.
A letter from Lutnick to Anthropic, cited by multiple outlets, noted that the company had coordinated with the government to proactively identify and address safety risks related to the models.
Anthropic has committed to informing the government if it detects any malicious activity involving the models. Lutnick’s letter also warned that the Trump administration reserves the right to reimpose restrictions in the future if circumstances change or if Anthropic does not meet its commitments.
White House Chief of Staff Susie Wiles expressed gratitude to industries working closely with the White House to implement the president’s executive order on promoting advanced AI innovation and safety.
He highlighted ongoing efforts around model access, guardrails, testing and security, and emphasized that the priority remains deploying the best technology as quickly as possible.
Which Claude model should users choose?
Anthropic’s paid tiers give users immediate access to Sonnet 5 through the Claude app, Claude Code, and the Anthropic API. Free tier users can also access the model.
For those who were temporarily locked out of Fable 5 and Mythos 5 during the two-week suspension, access will be gradually restored starting Wednesday.
Pro, Max, and Enterprise plan users should see the models reappear in their model selection menus.
When choosing a model, users can consider the following:
- Sonnet 5: Offers the best cost-performance balance for most tasks involving agency and coding.
- Opus 4.8: Provides maximum performance for the most demanding tasks.
- Fable 5: Shares architecture with Mythos and includes safety railings; It will be available at a usage-based price after the free period that ends June 22.
- Myths 5: Restricted to vetted partners, including government cyber defenders and certain life sciences researchers.
The ongoing events demonstrate Anthropic’s efforts to balance expanding its business with managing government regulations.
The launch of Fable 5 in June, its subsequent suspension, and now its reinstatement along with a more affordable Sonnet launch show how Anthropic is refining its product line as it grapples with increasingly complex rules for advanced AI models.
The launch of Sonnet 5 also strengthens Anthropic’s position against OpenAI. Recently, OpenAI delayed the public release of GPT-5.6 at the request of the US government, which limited access to select partners.
With Sonnet 5 now available and the return of Fable 5, Anthropic offers a broader range of models accessible to developers and users than OpenAI in the near future.
Sonnet 5 is now available. Access to Fable 5 and Mythos 5 will be restored starting Wednesday for eligible users.
