Inception Selected to Participate in the 2025 AWS Generative AI Accelerator

Inception, the company pioneering diffusion large language models (dLLMs), today announced it has been selected for the third cohort of the AWS Generative AI Accelerator. Launched by Amazon Web Services, Inc. (AWS), the AWS Generative AI Accelerator identifies top early-stage startups that are using generative AI to solve complex challenges and help them scale and grow. Participants will access AWS credits, mentorship, and learning resources to further their use of AI and ML technologies and grow their businesses.

By adapting diffusion technology proven in image and video generation to large language models, Inception is introducing dLLMs that generate blocks of text in parallel rather than sequentially. This shift delivers text generation that is 5-10x faster and more efficient, with enhanced reasoning capabilities and more precise control over output structure and quality.

This opportunity will support Inception’s efforts to advance dLLMs as faster, more efficient, and more controllable alternatives to traditional LLMs. Inception has built and deployed the first commercial-grade dLLM models, making it easier for enterprises and developers to deploy generative AI in real-world applications, from coding and customer support to complex reasoning tasks. By participating in the AWS Generative AI Accelerator, Inception will continue to advance its Mercury models on AWS services while broadening its visibility with customers, investors, and talent.

Inception’s foundation models, Mercury and Mercury Coder, are available on Amazon Bedrock Marketplace and Amazon SageMaker JumpStart. These models deliver ultra-fast generation speeds of up to 1,100 tokens per second on NVIDIA H100 GPUs, support long contexts of up to 128,000 tokens, and reduce GPU footprint, allowing enterprises to run larger models and serve more users with the same infrastructure. That efficiency makes the models well-suited for latency-sensitive applications like interactive voice agents, live code generation, and dynamic user interfaces. The availability of the Mercury family of models on Amazon Bedrock Marketplace and Amazon SageMaker JumpStart underscores Inception’s growing commercial readiness and contributions to diffusion-based language modeling.

Customers such as Radient, a Toronto-based AI research and development company, are using Mercury models to speed development workflows and lower inference costs.“We cut routing and classification overheads to sub-second latencies even on complex agent traces, which allowed us to build tools that make assistants feel fast while keeping the bills low for our customers,” said Radient founder Damian Tran.

“Inception was founded on the research demonstrating that diffusion-based architectures can unlock faster, more efficient, and more powerful generative AI,” said Inception CEO and co-founder Stefano Ermon. “Being selected for the AWS Generative AI Accelerator is both validation of our approach and a chance to scale our impact by working with one of the world’s leading cloud providers.”

All 40 global participating startups will be invited to attend and showcase their solutions to potential investors, customers, partners, and AWS leaders in December at re:Invent 2025 in Las Vegas.

“Whether it’s in biotech labs, creative studios, or industrial applications, the pace of generative AI innovation is extraordinary, and it’s happening everywhere,” said Sherry Karamdashti, General Manager & Head of Startups in North America at AWS. “This year’s cohort reinforces our mission to help that innovation move faster and deliver real-world impact for customers in every industry. We’re removing the barriers and accelerating opportunities so these leaders can grow their world-changing solutions.”

For more information on the Generative AI Accelerator, visit https://aws.amazon.com/startups/accelerators/generative-ai.

To try out Mercury, visit chat.inceptionlabs.ai.

To get started with the API, visit platform.inceptionlabs.ai.

About Inception

Inception is the pioneer of diffusion large language models (dLLMs), a new architecture that generates text in parallel to deliver 5-10x speed and efficiency gains over traditional autoregressive models for enterprise applications. Its flagship model, Mercury, is the world’s first and only commercially available dLLM. Founded by leading AI researchers from Stanford, UCLA, and Cornell, including diffusion co-inventor Stefano Ermon, the team combines deep academic expertise with experience from DeepMind, OpenAI, Meta, Microsoft, and NVIDIA to build the next generation of scalable, high-performance models.

For more information, visit www.inceptionlabs.ai.

Contact Information

Natalie Bartels

Media Contact

VSC, on behalf of Inception

inception@vsc.co