EL7.AI
StocksMediumUpdated×3•Originally published 9 November 2026•Updated 9 November 2026•
4 min read

Anthropic Launches Two Models 30% Faster and Reports $33.242 Billion in Revenue

Key Facts

1Anthropic said Fable 5.2 and Mythos 5.2 are up to 30% faster than their 5.1 versions, at $8 per million input tokens and $40 per million output tokens.
2Fable 5.2 scored 64.9% on Anthropic’s internal RoboDojo Sim test, compared with 27.6% for Fable 5.1.
3Anthropic revenue was $33.242 billion in the nine months ended September 30, 2026, compared with $2.545 billion in the corresponding period of 2025.
4Anthropic’s net loss was $104 billion, including a $94 billion accounting adjustment related to funding agreements.
5Total spending commitments are close to $895.4 billion, of which $69.8 billion is due in the 12 months following September 30, 2026.

Anthropic announced the launch of Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, saying they improve coding, reasoning and vision. The company said the models generate outputs up to 30% faster than their 5.1 versions, at a base price of $8 per million input tokens and $40 per million output tokens. The announcement puts the models’ performance and pricing at the center of the update, while access differs: Fable is more widely available, whereas Mythos is subject to verification programs for certain uses. Anthropic says both models are based on the same underlying model, and that Fable includes cybersecurity and biology safeguards to limit misuse. The announcement came as the company’s financial filing showed rapid revenue growth in the nine months ended September 30, 2026.

The company also offers an optional fast mode that increases token generation speed by up to 2.5 times, at $16 per million input tokens and $80 per million output tokens. That makes usage in this mode more expensive than at base pricing, in exchange for higher generation speed, according to Anthropic’s announcement. The available information does not cite different prices for Fable and Mythos; pricing was stated for both models. In Anthropic’s internal RoboDojo Sim test at the highest available effort, Fable 5.2 scored 64.9%. Anthropic compared that result with 39.4% for Opus 5.5, 29.8% for GPT-6 Astra and 27.6% for Fable 5.1 in the same test. The comparison remains tied to the company’s internal robotics simulation benchmark, and Anthropic presented the figures as part of its launch results.

In another early test, Fable 5.2 scored 45.3% at high effort on FrontierCode 2.0, compared with 45.5% for Opus 5.5 at maximum effort, according to Anthropic. The company said the task cost with Fable was about half its cost with the comparison model. Anthropic also said Fable 5.2 completed a formal proof of Fermat’s Last Theorem during 30 hours of pre-release testing. These results add evidence in coding and reasoning, areas the company said the releases improve. The achievements and results remain part of the tests Anthropic cited in its announcement, alongside the RoboDojo Sim score.

On availability, Fable 5.2 is now available on Anthropic’s cloud platforms, including Amazon Web Services, Google Cloud and Microsoft Azure. For Mythos, the company says organizations can apply for access through two verification programs covering cybersecurity and life sciences research, limiting its availability in those uses. Anthropic said Claude Security is available to Claude Enterprise customers and uses its model family, including Mythos 5.2. The arrangement separates a more widely available model with built-in safeguards from Mythos capabilities that require institutional verification in the areas cited. The company said Fable’s safeguards are intended to protect against misuse in cybersecurity and biology.

Anthropic’s filing with the U.S. Securities and Exchange Commission shows revenue of $33.242 billion in the nine months ended September 30, 2026, compared with $2.545 billion in the corresponding period of 2025. The Associated Press described the increase as about 13-fold, while net loss was $104 billion, including a $94 billion accounting adjustment related to funding agreements. On a quarterly basis, the agency said the company reported revenue of $17 billion and a net loss of $23.3 billion in the quarter ended September 30, 2026. In the same filing, the company reported an operating loss of $567 million in the three months ended on that date. Anthropic said the operating loss was partly affected by $5.0 billion in computing under short-term commitments and $2.0 billion in non-cash charges related to its donation-matching program.

Revenue growth has come alongside large spending commitments for computing and infrastructure. The filing says total committed spending is about $895.4 billion, with about $447.5 billion expected to be spent before December 31, 2030. About $69.8 billion is also due in the 12 months following September 30, 2026. The Associated Press said the company’s commitments include spending on computing and infrastructure facilities, such as data centers. These commitments put the scale of future spending in the context of the expansion reflected in revenue, while the financial data continue to show losses over the period covered.

On a potential public offering, the Associated Press said the updated filing did not specify how much Anthropic expects to raise or the price at which its shares would be offered. The company has previously said it intends to list its shares on the Nasdaq Global Select Market under the symbol ANT. The agency’s report did not include a confirmed offering date or details on the number or price of shares. The model launches come as the company expands its products, while the terms and potential proceeds of an offering remain unspecified in the published information cited by the sources. That leaves the timing and terms of the listing subject to future company announcements.