PRODUCT

Anthropic shifts the AI race toward finished work

Anthropic wants to pivot the conversation around AI from raw costs to how much work the models and agents can actually complete for the money spent.

On Monday, Anthropic released its most advanced frontier models, Fable 5.2 and Mythos 5.2, claiming that they are built to coordinate agents to orchestrate much more complex workflows and see them through to the finish. Anthropic claims that Fable 5.2 and Mythos 5.2 generate tokens 30% faster than Fable 5.1 and Mythos 5.1. The frontier lab, which is in the process of going public, also cut the price of both models by 20%. 

Fable can act as an Advisor, where less powerful (and less expensive) Anthropic models like Haiku, Sonnet, and Opus handle basic tasks but escalate harder problems to Fable. Conversely, Fable can operate in Orchestrator mode, delegating lower-level tasks to these smaller models. 

Fable 5.2 and Mythos 5.2 are still the same model under the hood, but Fable is the public-facing version with "strong cybersecurity and biology safeguards to guard against misuse," as Anthropic characterizes it. Meanwhile, Mythos is only available to verified "organizations pursuing beneficial applications in cybersecurity or biomedical research," Anthropic says. 

More to know about Fable 5.2 and Mythos 5.2:

• Both models are available for $8 per million input tokens and $40 per million output tokens; that's down from $10 input and $50 output and the same price as OpenAI's GPT-6 Astra (don't be surprised to see a price drop to match sometime soon). Google's Gemini 4 Carbon, one of the few models in the same league, costs significantly less at $4 per million input tokens and $20 per million output tokens (and it's 50% off that for an introductory period to entice more AI builders to switch).

• The two models are also launching with an optional fast mode that can generate tokens 2.5 times faster for twice the price ($16 per million input tokens and $80 per million output tokens).

• Fable 5.2 is available starting on Monday across AWS, Azure, and Google Cloud. Reputable firms can apply to get access to Mythos through the Cyber Verification Program or the Life Sciences Verification Program.

"You can just throw Claude Fable 5.2 at new, ambiguous systems it has never seen before, and it will figure out how to actually work within them," Matt Lee, senior staff engineer at Datadog, told Anthropic. "It's almost like an incident commander that can send out faster models to get quick signals and coordinate, which removes that burden from me as an operator."

Andrew Mager, staff engineer of agentic systems at Uber, reported, "I had Claude Fable 5.2 run a 10–12 hour security audit. It sent out about 170 reviewer agents and came back with a write-up, fix pull requests, and a policy proposal. It works until the job is actually complete instead of stopping at a plausible checkpoint."

With its leading-edge—but very expensive—models, Anthropic is trying to change the conversation around token economics to how much verified work a model can actually finish for the price of the tokens. The fact that Anthropic is building model routing into its platform is great for customers because it can potentially lower token costs and speed up tasks. It's also good for Anthropic because it means it won't get disintermediated by other routing platforms orchestrating between the best models, but instead allows the lab to offer tighter integration and coordination between its own models. At the end of the day, AI builders just want to get their work done and keep things fast and simple, while companies want to control AI and keep it within budget. At the same time, as professionals and companies accomplish more with AI, their appetite gets bigger for new work they can give it. That's where advanced models like Fable 5.2 and Mythos 5.2 are worth trying—more ambitious tasks that have failed in the past or newly ambitious ones that have emerged.

TOGETHER WITH STACKAI

Build No-Code AI Agents With Enterprise-Grade Governance

Anyone can build an AI agent. Almost nobody can get one past IT. StackAI is built for both.

Your builders get:

• Drag-and-drop across SharePoint, Slack, Teams, databases and 500+ systems

• Chat assistants, forms, Slack apps and API endpoints in minutes

Your IT team gets:

• Governed deployments with review and rollback

• Granular RBAC, SSO, audit logs and on-prem

Agents in production in weeks. Trusted by Morgan & Morgan, BAE Systems, City of Santa Monica, LifeMD + more.

Get your AI strategy today

GOVERNANCE

 Nvidia scandal exposes AI's accountability gap

Nvidia built an empire on AI, but its CEO's own actions may now be the company's biggest threat.

Over the weekend, confirmation of Nvidia CEO and co-founder Jensen Huang's extramarital affair was released as Jensen sent an internal message to the company apologizing for his actions, admitting to having been "unfaithful" to his wife and "deeply sorry." While that information alone would be enough to cause headlines, the woman involved is Alice Huang, who US intelligence has accused of being a Chinese spy. 

If you track the AI scene closely, you may already know that The New York Times broke the initial story about US intelligence investigations into the relationship at the end of October. Huang initially defended the relationship, telling Bloomberg News his dealings with Megaspeed International, where Alice Huang is a top executive, were "normal and not unusual." Jensen's admission in the internal memo has revived the story.

While many more details about the nature of the affair are in the in-depth New York Times article, the focus now turns to the aftermath, as the Nvidia board of directors issued a statement that goes as follows: 

• The board is "initiating an independent investigation in light of recent reporting that founder and CEO Jensen Huang has been the target of a foreign intelligence effort." 

• The board has "no reason to believe that either national security or Nvidia secrets were compromised."

• Jensen will cooperate with the review, which a committee of independent directors will conduct with help from outside independent counsel. 

• Nvidia has also stated that Jensen will continue to lead the company as CEO during the investigation, and that the board expressed its "confidence in Jensen and his leadership team." 

China has since denied engaging in attempts to spy on Nvidia's CEO, claiming the allegations are part of a smear campaign against Beijing. Lastly, as of Sunday, President Trump has defended Huang, telling reporters he would still take him on the trip to China and "I haven’t heard he’s done anything wrong…He dealt with somebody, but you know, so does everybody else." Despite both of these efforts to minimize the allegations, the markets have spoken, with Nvidia stock falling 5% as scrutiny mounts.

For investors, a company’s reputation is a key indicator of its health. So while a CEO’s affair is a personal matter, it can move markets, which is why The Deep View is covering it, even though we aren’t in the business of gossip. The details of this case also make it relevant to the wider AI market: US intelligence officials suspect the woman involved is a Chinese spy, and, reportedly, the Trump administration was aware of the concerns but chose not to act because of its close relationship with Huang. This makes the case for an independent regulatory body, just as the Hugging Face incidents did earlier this year. The AI race is now so contested that governments will go to great lengths to get ahead, whether that means recruiting spies to gather intelligence or ignoring espionage warnings for political reasons. An independent auditor and mandatory security-risk disclosures would make it harder for high-risk situations to slip through the cracks, which matters all the more as investors pour billions into the industry and national security depends on the models they are producing.

TOGETHER WITH GRANOLA

Our favourite AI notepad can now type for you

Our team already uses Granola to take notes in meetings. Now it can do the typing between them, too.

Our favourite use case for Voice Typing so far: AI prompts.

Instead of typing out a long prompt, press a hotkey and basically brain-dump everything you’re thinking. We’ve found it much easier to get all the context, ideas, and details out quickly, which gives the AI more to work with. That greatly improves the results and value you can get from your AI.

Other Voice Typing use cases we’ve found helpful:

• Voice type longer emails and messages

• Fire off Slack replies in quick succession

• Get thoughts or ideas down in notes quickly without stopping to type.

And because Granola already knows the names and jargon from your meetings, it has context for the way you and your team actually talk. Across our team, we haven’t had to correct a single word it’s transcribed yet, despite a few funky accents.

Try Granola Voice Typing, with unlimited words

PRODUCT

Meta’s Muse Ember wins on cost of agentic tasks

Meta has a new AI model that CFOs are going to love—and not because it automates spreadsheets.

On Monday, Meta launched Muse Ember, its budget-tier AI model to compete with Anthropic's Claude Haiku 5.5 and OpenAI's GPT-6. That's not to mention the latest budget-tier offerings from Google, Microsoft, and other frontier labs and neolabs. Muse Ember beats Haiku and Luna on lots of benchmarks and it looks like it will raise the bar on speed and performance per dollar.

The race for less expensive models that are capable enough to do most agentic tasks but can slow down ballooning AI bills has emerged as the hottest trend of the second half of 2026, as more and more organizations of all sizes find tasks to streamline with AI. 

Simultaneously, it's no secret that AI costs have been scaling out of control for most of the year, with AI agents being one of the main culprits. As a result, CFOs and other leaders have made costly, inefficient use of AI models one of their biggest targets for reigning in spending. That's why there are new announcements almost every day about model routing and cheaper, more budget-friendly models. 

Here's what Meta's Muse Ember brings to the table:

• Ember matches Haiku and Luna in base cost at $0.10 per million input tokens and $0.50 per million output tokens, but beats them in cached input per million tokens as they both come in at $0.01 and Ember costs $0.004 (60% less); as a result, Artificial Analysis estimates the approximate cost per task as $0.14 for Ember, $0.15 for Luna, and $0.41 for Haiku.

• In the AutomationBench-AA for business workflows, Ember scored 37.8% compared to 13.9% for Haiku 5.5 and 20.2% for GPT-6 Luna

• In the Coding Agent Index-AA benchmark, Ember scored a 46, while Haiku 5.5 scored 37 and GPT-6 Luna scored 31.

"What stood out in our testing was speed. Muse Ember ran up to 30% faster at the tail than the comparable model we tested, and it handled broad, multi-repo investigation tasks well. It’s a strong fit for fast, high-throughput agentic workloads," said Cloudflare CTO Dane Knecht, as quoted in a Meta statement. 

Of course, models leapfrog each other day. Who knows how long it will be before other budget-tier models leapfrog Muse Ember, or which ones it will be. But this is another indication that Meta is catching up to the leading frontier labs to drive more competition, and that models in general are commoditizing. At the end of the day, the question for leaders and AI builders is how much useful work can a fixed AI budget buy? And the answer is more every day, if you use the right models and route your workloads effectively. Muse Ember is laser-focused on making the cost of long-running, routine tasks by agents a lot cheaper. And based on the early benchmarks, it's about as powerful as Anthropic's Haiku and even a little cheaper than OpenAI's Luna. 

LINKS

• Adobe and Disney forge multi-year AI tools alliance

• Meta expands AI age detection to put more teens in protected accounts

• Medium launches TK, an AI-powered note-taking and private social app

• Nominal Inc. raises $240 million for manufacturing efficiency software

• Boston Dynamics Atlas performs 360-degree jumps and wheel-based maneuvers

• ByteDance eyes expanding a data center in Brazil 

• Seedream 5.5: ByteDance made new image-generating model available globally

• ChatGPT Dots: Can use sign-in codes sent to your connected inbox to log in for you

• ElevenCreative: Topaz upscaling controls are now on the platform 

• Google Gemini: Rolling out ‘Memory summary’ with everything it remembers about you

• Greylock Partners: AI Researcher, Agent Systems

• Obin AI: Agentic AI Engineer

• Atlassian: Customer Success Architect (AI)

• DeepRec.ai: Machine Learning Researcher, Agentic RL

GAMES

POLL RESULTS

Do you think that physical AI use cases will scale significantly by 2030?

Yes (56%)
Somewhat (32%)
No (10%)