Introducing Qwen3: Hybrid Thinking Modes and Advanced AI Now in Enclave AI

We are thrilled to announce the integration of the cutting-edge Qwen3 family of large language models into Enclave AI. This update brings powerful new capabilities, including innovative Mixture-of-Experts (MoE) architectures and unique hybrid thinking modes, directly to your device while upholding our core commitment to absolute privacy.

Advanced AI with Hybrid Thinking

Qwen3 represents a significant leap forward in AI development, offering models that balance performance, efficiency, and flexibility. Key highlights include:

• Mixture-of-Experts (MoE) Architecture: Models like Qwen3-30B-A3B deliver exceptional performance comparable to much larger models while activating only a fraction of their parameters (3 billion out of 30 billion total), making them ideal for efficient on-device processing.
• Hybrid Thinking Modes: Qwen3 introduces a novel approach allowing users to switch between:
  • Thinking Mode: The model reasons step-by-step for complex problems, providing deeper analysis.
  • Non-Thinking Mode: The model offers near-instant responses for quicker tasks. This flexibility, controllable within Enclave AI, allows you to tailor the AI’s approach to your specific needs, balancing depth and speed.
• Competitive Performance: Qwen3 models demonstrate strong performance across benchmarks in coding, math, and general capabilities, rivaling top-tier models. Even smaller Qwen3 models show remarkable proficiency.
• Extensive Multilingual Support: With support for 119 languages and dialects, Qwen3 significantly expands Enclave AI’s usability for users worldwide.

What Qwen3 Means for Enclave AI Users

Integrating Qwen3 brings tangible advantages to your private AI experience:

Enhanced Problem Solving: Leverage the “Thinking Mode” for complex tasks requiring detailed reasoning, all performed securely on your device. Faster Interactions: Switch to “Non-Thinking Mode” for rapid answers to simpler queries without sacrificing privacy. Greater Efficiency: Benefit from the power of MoE models that deliver high-end performance without excessive resource drain on your device. Broader Language Access: Interact with Enclave AI in a wider range of languages.

Availability in Enclave AI

We are initially rolling out support for select Qwen3 models on iOS, with macOS support coming soon. We’ve optimized these models for the best possible on-device performance within Enclave AI’s secure environment. iOS users can choose Qwen3 from the model selection menu, just like any other model.

Privacy Remains Paramount

As always, our integration of Qwen3 adheres strictly to Enclave AI’s privacy principles:

• All processing happens entirely on your device.
• No data is ever sent to external servers.
• Your interactions remain completely private and offline.

Explore the Power of Qwen3 Today

Update your Enclave AI app now to access the Qwen3 models. Experiment with the hybrid thinking modes and experience the next level of private, on-device AI. We believe Qwen3’s unique capabilities offer a powerful new dimension to secure, personalized AI assistance.

We look forward to hearing your feedback as you explore Qwen3 within Enclave AI. Your insights drive our mission to deliver the most advanced and private AI tools possible.

Reference: Qwen3: Think Deeper, Act Faster - Qwen Blog