Anthropic has launched Claude Fable 5.2 and Claude Mythos 5.2, the fastest frontier AI models it has ever released, just days before one of the largest initial public offerings in stock market history. The two models, announced on Monday, November 9, 2026, generate outputs up to 30 percent faster than their predecessors and cost 20 percent less to use.

Key Points

• Anthropic launched Claude Fable 5.2 and Claude Mythos 5.2 on Monday, November 9, 2026, ahead of an IPO expected the week of November 16 that could raise up to $100 billion.
• Both models generate outputs up to 30 percent faster than Fable 5.1 and Mythos 5.1, with prices cut 20 percent to $8 per million input tokens and $40 per million output tokens.
• Fable 5.2 scored 49.1 percent on the FrontierCode 2.0 benchmark, up from 32.4 percent for Fable 5.1, the highest score among released models Anthropic has tested.
• In a demonstration, Fable 5.2 coordinated more than 1,000 AI agents over eight days to build a working operating system prototype.
• Mythos 5.2 identified 89.6 percent of vulnerabilities in Anthropic's internal CVE Rediscovery evaluation, up from 78.3 percent for Mythos 5.1, but remains restricted to vetted organisations.

What Happened

Anthropic announced the two models on Monday, November 9, 2026, describing them as updated versions of its two most powerful AI systems. Both share the same underlying model but serve different jobs: Fable is the general-purpose flagship for coding, research and business, while Mythos is built for cybersecurity engineering and is kept behind a vetting process.

The headline improvement is speed. Both models generate outputs up to 30 percent faster than Fable 5.1 and Mythos 5.1, which matters most for long, multi-step tasks. Anthropic also added an optional fast mode, up to 2.5 times faster at higher pricing tiers.

Prices are coming down too. The new models cost $8 per million input tokens and $40 per million output tokens, down from $10 and $50, a cut of 20 percent. Fable 5.2 is available now on Anthropic's platforms and through Amazon Web Services, Google Cloud and Microsoft Azure. Anthropic still recommends Claude Opus 5.5 for most everyday coding and knowledge work.

Background and Context

Anthropic, founded in 2021 by former OpenAI researchers, builds Claude, one of the leading large language models, and sells it to businesses and developers. The company filed to go public in October 2026, and the listing is the most closely watched test yet of investor appetite for pure-play AI companies.

Anthropic began its IPO roadshow on Tuesday, November 10, 2026, a day after emailing bankers a $70-to-$80-a-share price range and then retracting it. A retracted filing copy showed plans to sell 867 million shares for up to $69.3 billion, implying a valuation of about $1.4 trillion, with trading expected the following week on the Nasdaq under the ticker ANT, before Thanksgiving on November 26.

The filings show why: roughly $17 billion in third-quarter 2026 revenue, up about 1,127 percent year on year, but a $23.29 billion net loss and $895.4 billion in compute commitments stretching to 2045.

Claude Fable 5.2: Faster Coding and a Record Benchmark Score

The strongest evidence is its score on FrontierCode 2.0, a benchmark for difficult coding tasks. At the highest effort setting, Fable 5.2 scored 49.1 percent, up from 32.4 percent for Fable 5.1, the highest score among released models Anthropic has tested. It also outperformed Claude Opus 5.5 across benchmarks covering agentic coding, scientific research, business workflows, cybersecurity, robotics and visual understanding.

Anthropic backed the numbers with a striking demonstration. In one test, Fable 5.2 coordinated more than 1,000 AI agents over eight days to build a working operating system prototype. The model was also used to design three working electronic devices: a home voice-assistant hub, a USB speakerphone and a handheld virtual pet. It shows frontier models moving beyond short answers and code snippets, towards running long projects on their own.

Reactions and Quotes

Anthropic framed the launch around both capability and caution. Fable 5.2 carries enhanced cybersecurity and biology safeguards so it can be broadly available, the company said, while Mythos 5.2 is restricted to vetted organisations through its Cyber Verification Program. The split is deliberate.

Mythos 5.2 identified 89.6 percent of vulnerabilities in Anthropic's internal CVE Rediscovery evaluation, up from 78.3 percent for Mythos 5.1. That large jump explains the restriction. A model that good at finding weaknesses could attack systems as easily as defend them, which is why Anthropic keeps it behind a vetting process.

Why It Matters

The launch raises the bar for the whole industry. A 49.1 percent FrontierCode 2.0 score and a working operating system built by more than 1,000 agents set a new standard for OpenAI, Google and Meta. The price cut adds pressure too: a faster, cheaper model changes the economics for businesses choosing an AI provider.

The timing ties the technology directly to the money. Anthropic is asking investors to value it at around $1.4 trillion, which would make it one of the most valuable companies ever to go public. A launch that lands well strengthens that case; one that disappoints gives sceptics another reason to push the price down. The $895.4 billion in compute commitments shows how expensive the race is.

What Happens Next

Fable 5.2 is available now, and developers and businesses will start testing it against their own workloads in the coming days. The real verdict will come from whether the benchmark gains translate into faster, cheaper work on real projects, not only on Anthropic's own evaluations. Mythos 5.2 will roll out more slowly through the Cyber Verification Program.

Attention now turns to the IPO. Investor meetings run this week, with Anthropic expected to announce its corrected price range in the coming days, aiming for trading to start on Thursday or Friday of next week, November 19 or 20, on the Nasdaq under the ticker ANT. A range near the retracted $70 to $80 would signal the market can still support a valuation of around $1.4 trillion.

Frequently Asked Questions

What are Claude Fable 5.2 and Mythos 5.2? They are Anthropic's newest AI models, announced on Monday, November 9, 2026: Fable 5.2 for coding, research and business work, and Mythos 5.2 for cybersecurity engineering.

How much faster are the new models? Both generate outputs up to 30 percent faster than Fable 5.1 and Mythos 5.1, and an optional fast mode runs up to 2.5 times faster at higher pricing tiers.

How much do Fable 5.2 and Mythos 5.2 cost? Prices were cut 20 percent to $8 per million input tokens and $40 per million output tokens, down from $10 and $50.

Why is Mythos 5.2 restricted? It is Anthropic's most capable cybersecurity model, available only to vetted organisations through its Cyber Verification Program, because a model that good at finding vulnerabilities could also be misused.

When does Anthropic's IPO happen? The roadshow began on Tuesday, November 10, 2026, and trading is expected to start on the Nasdaq under the ticker ANT the week of November 16, in a deal that could raise up to $100 billion.

Read more on US News Pulse: Anthropic IPO roadshow begins, Anthropic IPO valuation may drop and AI disproves Exponential Time Hypothesis.