Brief IA

360 and Sakana AI Challenge Anthropic with Innovative AI Models

💼 Business & Startups·Tom Levy·

360 and Sakana AI Challenge Anthropic with Innovative AI Models

360 and Sakana AI Challenge Anthropic with Innovative AI Models
Key Takeaways
1The Chinese company 360 has launched Tulongfeng, an AI model competing with Anthropic's Mythos, despite U.S. bans.
2Tokyo-based Sakana AI has unveiled Fugu, an AI model designed to compete with Anthropic's Fable 5 and Mythos Preview.
3U.S. export bans are driving AI innovation in Asia, with startups filling the gap left by Anthropic.
💡Why it mattersThese developments highlight the rise of Asian AI, challenging U.S. dominance and redefining the global tech landscape.
Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisis ton rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

📄
Full Analysis

360 and Sakana AI Launch Innovative AI Models Amid U.S. Restrictions

On Wednesday, the Chinese cybersecurity company 360 unveiled its latest artificial intelligence model, Tulongfeng. This model is positioned as a direct competitor to Mythos, developed by Anthropic. The latter, due to its power and cybersecurity implications, is currently subject to an export ban imposed by the Trump administration, similar to its more limited version, Fable 5. This ban aims to restrict access to these technologies for non-Americans.

At the same time, Sakana AI, a Tokyo-based startup, introduced Fugu, an AI model named after the Japanese pufferfish. According to Sakana AI, Fugu is designed to compete with cutting-edge models such as Fable 5 and Anthropic's Mythos Preview. This model is particularly innovative as it allows for orchestrating access to other models via their APIs, making it highly flexible and adaptable.

Context of U.S. Bans

These Asian launches come at a time when the U.S. government has recently extended a ban preventing Anthropic from exporting its Mythos and Fable models. This decision, made two weeks ago, aims to protect national interests in cybersecurity, but it has also opened the door for Asian players to fill the gap left by these restrictions.

A spokesperson for Sakana AI explained to TechCrunch that the launch of Fugu was coincidental in relation to the U.S. ban, although the company has been able to leverage it. On their website, Sakana AI highlights Fugu's ability to deliver top-tier performance without being subject to U.S. export controls.

Sakana AI: A Thoughtful Strategy

Founded in 2023 by former Google researchers, including Ren Ito, Llion Jones, and David Ha, Sakana AI focuses on developing accessible generative AI models optimized to work with small datasets and tailored to Japanese language and culture. While the company primarily targets Japanese businesses and government agencies, it does not claim that this marks a lasting shift away from American models.

The spokesperson for Sakana AI emphasized that American models remain crucial for Asia, a position echoed by Ren Ito at the G7 summit in Évian. Ito advocated for the U.S. to preserve access to AI for its allies, asserting that AI should not be monopolized but developed collaboratively.

A Broader Vision for AI

David Ha, CEO of Sakana, described Fugu as an innovation that goes beyond opportunism in light of American difficulties. Fugu is designed to coordinate the use of agents among various models, representing a significant advancement over traditional models. Ha expressed on X that relying on a single provider for national infrastructure is risky, a lesson reinforced by recent export controls.

He highlighted that access to the best models can be revoked overnight, and that collective intelligence is a practical solution to counter this concentration of power.

360's Response

In contrast to Sakana, the Chinese company 360 takes a more direct approach. It has launched two AI security tools: Tulongfeng, which automatically detects software vulnerabilities, and Yitianzhen, which automates cybersecurity defense and incident response. The founder of 360, Zhou Hongyi, described vulnerability detection AI as a national strategic asset, warning against the risk of "one-way transparency" where certain actors would have exclusive access to these advanced capabilities.

Implications for Anthropic

Anthropic has experienced impressive growth, reporting an annual revenue of $47 billion in May 2026. However, the reliance of this success on Asian clients remains uncertain. Since the export ban came into effect, companies in Tokyo and Beijing have begun to occupy the space left vacant. Even if American companies can regain trust once the ban is lifted, local alternatives, better suited to linguistic and cultural specifics, are already in place.

360 has not responded to requests for comments, but it is clear that Asian AI players are ready to seize the opportunities created by U.S. restrictions.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.