What is Claude Fable 5?

Claude Fable 5 is Anthropic's most capable AI model, released initially in early 2026. It features a 1M-token context window, state-of-the-art performance on FrontierCode, CursorBench, and frontier physics and finance benchmarks. It costs $10 per million input tokens and $50 per million output tokens — double the price of Opus 4.8. Fable 5 leads on the hardest reasoning and coding tasks.

Why was Claude Fable 5 removed from some markets?

Fable 5 was affected by US export controls on advanced AI model weights. The controls restricted access to certain frontier AI capabilities for security reasons. Anthropic complied with these regulations, which limited Fable 5 availability. After a policy review, the export controls were lifted, allowing full redeployment.

What is the new cybersecurity classifier in Fable 5?

The cybersecurity classifier is a dedicated safety layer that blocks reported jailbreaks and prompt injection attacks specifically targeting Fable 5. It was added during the redeployment and is designed to detect and prevent attempts to bypass the model's safety training. Anthropic says the classifier addresses vulnerabilities identified during the export control review period.

How does Fable 5 compare to Opus 4.8 and GPT-5.6 Sol?

Fable 5 is Anthropic's flagship at $10/$50 per MTok, outperforming Opus 4.8 ($5/$25) on the hardest benchmarks. Its lead over Opus 4.8 widens the longer and more complex the task. Compared to GPT-5.6 Sol ($5/$30), Fable 5 is stronger on coding and reasoning benchmarks but more expensive. Fable 5's 1M-token context window is significantly larger than GPT-5.6 Sol's ~400K practical limit.

Who can access Claude Fable 5?

Fable 5 is available through the Claude API, Amazon Bedrock, and via Claude Pro/Max subscription tiers (Max 20x required for full access). It was redeployed on July 1, 2026 after the export controls were lifted. Access may still be restricted in certain countries subject to US AI export regulations. API users can access Fable 5 through the claude-fable-5 model endpoint.

Quick Answer

Claude Fable 5 Redeployed After US Export Controls Lift: July 2026 Update

Published: July 5, 2026

Claude Fable 5 Redeployed After US Export Controls Lift: July 2026 Update

Anthropic redeployed Claude Fable 5 on July 1, 2026, following the lifting of US export controls that had restricted access to the company’s most capable AI model. The redeployment includes a new cybersecurity classifier designed to block jailbreak attempts.

Fable 5 is Anthropic’s frontier model — significantly more capable than the already-powerful Opus 4.8, but also more expensive at $10/$50 per million tokens. Its 1M-token context window makes it the best Anthropic option for tasks requiring sustained, complex analysis across large codebases or documents.

What Changed with the Redeployment

Before (export controls active)	After (July 1, 2026+)
Fable 5 access restricted or unavailable in some markets	Full redeployment across API, Bedrock, and consumer tiers
No dedicated cybersecurity classifier	New jailbreak-blocking classifier added
Limited testing of Fable 5’s vulnerabilities	Classifier based on findings from export-control review period
Uncertainty about future availability	Clear regulatory path for continued deployment

The cybersecurity classifier is specifically notable. Anthropic says it targets “reported jailbreaks” and prompt injection attacks that attempt to bypass Fable 5’s safety training. This suggests the export control review period may have identified specific attack vectors that the classifier now addresses.

Fable 5 Pricing and Positioning

	Claude Fable 5	Claude Opus 4.8	Claude Sonnet 5
Input price/MTok	$10.00	$5.00	$3.00 ($2.00 intro through Aug 31)
Output price/MTok	$50.00	$25.00	$15.00 ($10.00 intro through Aug 31)
Context window	1M tokens	200K tokens	200K tokens
Fast mode	Not available	✅ 2.5x at $10/$50	Not available
Best for	Hardest frontier tasks	Daily agentic work	Cost-efficient agents
Model access tier	Max 20x / API	Pro+ / API	Pro / API

Fable 5 is 2x the price of Opus 4.8 but with a substantially larger context window and better performance on the hardest benchmarks. For most development work, Sonnet 5 or Opus 4.8 provide better value — Fable 5 is for the tasks where only the absolute best will do.

Benchmark Performance

Fable 5 leads Anthropic’s lineup on every major benchmark:

Benchmark	Fable 5	Opus 4.8	Sonnet 5
FrontierCode	SOTA	Strong	Very Good
CursorBench	SOTA	Strong	Very Good
Humanity’s Last Exam	~22%	~18%	~13%
SWE-bench Verified	~64%	~58%	~49%

Fable 5’s lead widens as task complexity increases. On the simplest queries, the gap between Fable 5 and Opus 4.8 or even Sonnet 5 narrows significantly.

Should You Use Fable 5?

Use Fable 5 when:

The task genuinely requires the maximum capability available (e.g., novel research, hardest debugging)
You need the 1M context window for large-codebase analysis
You’ve already determined Opus 4.8 or Sonnet 5 can’t handle the task
Quality is paramount and cost is secondary

Don’t use Fable 5 when:

The task is straightforward and a cheaper model can handle it
You’re cost-optimizing a high-volume agentic pipeline
You only need standard context lengths (200K or less)
Fast response time is critical (no Fast Mode for Fable 5)

The Bottom Line

Claude Fable 5’s redeployment with the new cybersecurity classifier is good news for teams that need Anthropic’s maximum capability. The export control episode shows that frontier AI models face increasing regulatory scrutiny — and that companies like Anthropic are building security improvements into the compliance process.

For most practical purposes, Opus 4.8 or Sonnet 5 remain the right daily drivers. Fable 5 is the heavy artillery: expensive, powerful, and best reserved for the tasks where only the best model will do.

Published July 5, 2026. Source: Anthropic (MarkTechPost July 1 coverage, Claude platform documentation). Export controls and cybersecurity classifier details are based on Anthropic’s public announcements.