Anthropic released Claude Sonnet 5 on June 30, 2026, positioning it as the default model for free and paid users at introductory pricing of $2 per million input tokens and $10 per million output tokens through August 31. The model performs nearly as well as Anthropic's more expensive Opus 4.8 on coding, reasoning, and computer-use tasks, but there's a critical context that changes what this release actually means: Anthropic's two most capable models, Fable 5 and Mythos 5, have been offline under a U.S. government export-control order since June 12, with no confirmed return date.

Why Is Sonnet 5 Suddenly the Best Claude Model You Can Use?

Until the export restrictions are lifted, Sonnet 5 is not just an incremental upgrade. For nearly everyone with a Claude subscription or API access, it is the highest-capability model available. Opus 4.8 remains fully accessible, but Sonnet 5 narrows the performance gap at a fraction of the cost, making it the practical choice for most users.

On June 12, the U.S. Commerce Department sent Anthropic an export-control directive ordering it to suspend access to Fable 5 and Mythos 5 for any foreign national, anywhere in the world. The government cited national security concerns and a reported technique for bypassing Fable 5's safeguards, though it did not provide detailed technical justification. Anthropic responded by calling the underlying jailbreak narrow and non-universal, noting that similar capabilities exist in other publicly deployed models. The company disabled both models for all customers worldwide within hours to comply with the order.

How Does Sonnet 5 Actually Perform Against Anthropic's Other Models?

Anthropic published benchmark results showing Sonnet 5 landing between Sonnet 4.6 and Opus 4.8 on most tasks, but notably closer to Opus. On one benchmark, Sonnet 5 even edges ahead of Opus 4.8, marking the first time a mid-tier Sonnet model has outperformed its full-size sibling.

• Agentic Coding (SWE-bench Pro): Sonnet 5 scores 63.2%, compared to 69.2% for Opus 4.8, demonstrating strong performance on real-world software engineering tasks.
• Reasoning Without Tools (Humanity's Last Exam): Sonnet 5 handles graduate-level reasoning across disciplines, landing between Sonnet 4.6 and Opus 4.8 on this benchmark.
• Computer Use (OSWorld-Verified): Sonnet 5 shows solid capability at operating real computer interfaces, a key metric for autonomous agent work.
• Knowledge Work (GDPval-AA v2): Sonnet 5 actually scores slightly higher than Opus 4.8, a narrow gap likely within methodology noise but notable since it is the cheaper model.

Sonnet 5 carries a 1 million token context window, meaning it can process roughly 750,000 words at once, and supports a 128,000 token output limit, the same as Sonnet 4.6. It runs on an updated tokenizer that processes the same input as roughly 1.0 to 1.35 times as many tokens depending on content type. Anthropic set introductory pricing to make that transition roughly cost-neutral through August 31, but standard pricing will kick in on September 1.

What's the Pricing Story, and When Does It Change?

The introductory rate is aggressive. Through August 31, Sonnet 5 costs $2 per million input tokens and $10 per million output tokens. After that date, it settles at $3 per million input tokens and $15 per million output tokens, which is exactly what Sonnet 4.6 already costs. This undercuts Opus 4.8 by 40% on both input and output pricing.

For context, Fable 5, the model that is currently suspended, previously cost $3 per million input tokens and $15 per million output tokens. The government cleared a more restricted version of Mythos 5 for a small set of vetted critical-infrastructure organizations on June 26, but that clearance did not extend to Fable 5, the version anyone with a subscription or API key was actually using.

How to Migrate to Sonnet 5 Without Breaking Your Integration

• Extended Thinking Behavior: Adaptive thinking is now on by default, and manual extended-thinking controls that existed on Sonnet 4.6 are gone. Requests that try to set a thinking budget directly will return an error, so you need to retest integrations rather than assuming a drop-in swap.
• Tokenizer Changes: The updated tokenizer means the same input produces 1.0 to 1.35 times as many tokens. Anthropic set introductory pricing to offset this, but standard pricing begins September 1, so budget time to recalculate costs for your use case.
• Safety and Cyber Safeguards: Sonnet 5 hallucinates and engages in sycophantic behavior less often than Sonnet 4.6, and resists prompt-injection hijack attempts more effectively. Cyber safeguards are on by default, the same detection system used on Opus 4.7 and 4.8.
• Security Research Limitations: Anthropic deliberately did not train Sonnet 5 on cybersecurity tasks. In a published test with Mozilla that asked models to develop working exploits for already-patched Firefox vulnerabilities, Sonnet 5 never produced a complete exploit. For security research requiring reduced guardrails, Anthropic recommends Opus 4.8 instead.

One feature worth noting: Priority Tier is not available on Sonnet 5. Anthropic also raised rate limits across Chat, Cowork, Claude Code, and the Claude Platform API to cover the heavier token usage of higher effort levels.

What Happens When the Export Controls Are Lifted?

Anthropic says discussions with the administration that could restore access to Fable 5 and Mythos 5 include the Sonnet 5 release itself, though no date is confirmed and earlier reports of an imminent return have already been retracted once. The timing carries significant stakes: Anthropic has confidentially filed for a U.S. public listing, and a recent funding round valued the company near $965 billion, so how quickly this gets resolved is being watched well beyond Anthropic's own customers.

Until that changes, Sonnet 5 is the practical ceiling for most users. It represents a genuine capability jump at a price point that makes it accessible to developers and enterprises that might have found Opus 4.8 too expensive. For Anthropic's paying customers, that's a meaningful shift in what the company's product lineup actually offers in the real world.