Ultra-fast next-edit prediction for coding
Mercury,from Inception Labs, is the first commercial diffusion LLM. Up to 10x faster than autoregressive models, with comparable or better quality on coding tasks.
Mercury Edit 2 is a small, coding-focused diffusion LLM designed for extremely latency-sensitive code editing workflows. It generates tokens in parallel rather than sequentially, enabling instant code completions and intelligent tab suggestions that keep developers in flow without interrupting their thinking process.
Mercury Edit 2 delivers 5x faster inference speeds and costs less than half of traditional auto-regressive LLMs. At $0.25 per 1M input tokens and $0.75 per 1M output tokens, it enables cost-effective scaling of coding assistants and real-time AI features in development environments.
Mercury Edit 2 is available through AWS Bedrock and Azure Foundry, providing enterprise-grade deployment options with 99.5%+ uptime and priority support. It supports fine-tuning and private deployments for organizations requiring customized coding assistance.
Mercury Edit 2's parallel token generation and diffusion-based architecture make it ideal for real-time code editing scenarios where latency is critical. It integrates as an OpenAI API-compatible drop-in replacement, enabling rapid deployment into existing coding tools and workflows.