PrismML Launches World's First 1-Bit AI Model Bringing Fast AI to Edge Devices

Pasadena, Calif. — February 11, 2026 — Prismatic, a pioneer in high-performance AI models for the edge, today announced its official launch and the release of its flagship 1-bit BonsAI 8B model. BonsAI 8B is the world’s first production-ready 1-bit AI model, delivering datacenter-class intelligence on phones, laptops, and other edge devices. Prism is building a future where powerful AI runs locally, efficiently, and securely — without massive data centers or massive energy costs.

With inference costs now estimated to be as much as 25x higher than training costs, and major tech companies spending $10–15 billion annually on model operations, Prism enables modern AI to run locally and cost-effectively on everyday devices. This unlocks new possibilities for real-time, privacy-first applications that were previously impractical at the edge. Prism’s technology could offer potential industry-wide savings of more than $10 billion per year.

The 1-bit BonsAI 8B model uses a proprietary Caltech mathematical breakthrough to outperform leading full-precision models such as Llama 8B and Marin 8B (placeholders), while being 16x smaller, 8x faster, and 4.5x more energy efficient. Despite its dramatically smaller footprint, BonsAI 8B achieves a benchmark score of 72 — nearly matching the 74 score of state-of-the-art, full-precision 8B models. The model brings powerful AI out of centralized data centers and into real-world environments where latency, power, and privacy matter most.

Download 1-bit BonsAI 8B here
Read the whitepaper here

“We're not just making AI more efficient — we're fundamentally democratizing who can build and deploy it,” said Dr. Babak Hassibi, CEO and founder of Prism and a professor at the California Institute of Technology. “We spent years developing the rigorous mathematics required to compress a neural network to a single bit without losing its reasoning capabilities. This shift democratizes AI, enabling life-changing applications in healthcare, defense, and finance to process sensitive data locally on-device, ensuring privacy and reducing reliance on scarce, high-performance chips.”

“AI’s future will not be defined by who can build the largest data centers,” said Vinod Khosla, founder of Khosla Ventures. “Prism's approach doesn’t just reduce costs — it unlocks entirely new categories of AI applications that were previously impossible due to power, latency, and privacy constraints. This is the kind of non-consensus innovation that creates the next generation of Fortune 500 companies.”

1-bit BonsAI 8B model benefits

• Radical efficiency: Runs 8x faster while using 10x less memory and 4.5x less energy than traditional models.
• Universal deployment: Brings datacenter-class AI to devices with 1/16th the footprint, enabling AI anywhere from smartphones to IoT sensors.
• Real-time performance: Eliminates cloud latency for split-second decisions in robotics, autonomous systems, and edge applications.
• Privacy by design: Keeps sensitive data on-device instead of sending it to tech giants — critical for healthcare, finance, and defense.
• Democratic access: Empowers smaller companies and nations to compete without billion-dollar budgets or scarce high-performance chips.
• Proven science: Built on proprietary Caltech IP solving a 40-year research problem through rigorous mathematics — not expensive trial-and-error.

Technical details

The 1-bit BonsAI 8B model is an 8-billion parameter large language model quantized to 1-bit precision. It is designed for seamless integration with existing AI workflows and optimized for low-latency inference on consumer-grade CPUs, NPUs, and edge GPUs. The model delivers high-fidelity reasoning and language understanding comparable to FP16 (16-bit floating point) models, but with a fraction of the memory footprint.

• Model precision: 1-bit
• Performance: 8x faster, 10x less memory, 4.5x less energy
• Benchmark score: 72 vs. 74 for state-of-the-art full-precision 8B models (97% of full-precision performance)
• Comparisons: Outperforms leading 8B models such as Llama 8B and Marin 8B on standard benchmarks (placeholders)
• Compatibility: Runs on phones, laptops, and edge devices

Pricing and availability

Developers and researchers can download the BonsAI 8B-1bit model for free starting today.

Download here

About Prism

Prism is a U.S.-based artificial intelligence company enabling powerful models to run locally on any device — ensuring privacy, reducing energy consumption, and eliminating reliance on massive data centers. Prism is built on proprietary Caltech intellectual property and backed by Caltech, Google, and Khosla Ventures. For more information, visit the company website, LinkedIn, or X.

All registered trademarks and product identifiers belong to their respective corporate entities. Any other trademarks or product names referenced here are also owned exclusively by their respective owners.

Media contact
Gary Bird
43PR
Prism@43pr.com
831.888.9011