Mercury 3 launches on Venice, enabling parallel token generation

Venice: Venice is a platform that hosts advanced AI models including Mercury 3 under an anonymized privacy tier. It allows permissionless access with prompts neither stored nor used for training and no personal generation history linked to accounts. Users pay per token via the Venice app or API without requiring a subscription. Mercury 3: Mercury 3 is Inception Labs' diffusion-based language model that generates and refines text in parallel across its full output rather than writing left to right. Released in October 2026, it supports a large context window and targets high-speed applications for agents and search. The model is available anonymously on the Venice platform through both the app and an OpenAI-compatible API. Inception Labs: Inception Labs develops advanced AI language models, focusing on diffusion-based architectures that enable parallel token generation instead of sequential processing. The company released its flagship model Mercury 3 in October 2026 as a solution for latency-sensitive uses including agents, search, and real-time applications. Mercury 3 is hosted on the Venice platform for anonymous user access. Access Model: Mercury 3 runs on Venice under an anonymized privacy tier where prompts are not stored, profiled, or used for training. Model Architecture: Mercury 3 employs a diffusion model to produce and refine answers in parallel rather than sequentially.