Inception Labs · od Februar 2025 · 5× · naposledy 29. 6. 2026
4
Momentum
Mercury is a family of commercial-scale diffusion-based large language models from Inception Labs that generates text through parallel iterative refinement rather than sequential token prediction. The model achieves over 1,000 tokens per second on NVIDIA H100 GPUs and is up to 10 times faster than comparable autoregressive models while maintaining comparable quality. Mercury is provided via API, on-premise deployments, and a web interface.
Vývoj momenta
19.05.17.08.
Vlastnosti
| Base Model | Proprietary diffusion transformer architecture (dLLM), not autoregressive; parameter count undisclosed |
| License | Proprietary commercial model (API access, not open source) |
| Platform | Inception API Platform, AWS Bedrock, SageMaker JumpStart, Azure AI Foundry, OpenRouter |
| Price | Mercury 2: $0.25 / 1M input tokens, $0.75 / 1M output tokens (Mercury 1: $0.25 / $1.00); 10M free tokens on account creation |
| Release Date | February 2025 (Mercury/Mercury Coder); Mercury 2 on March 4, 2026 |
| Interface (IDE/CLI/Web) | OpenAI-compatible REST API, chat web interface (Mercury Chat), IDE integrations (e.g. Zed, Continue, VSCode) |
| Supported Languages | Python, JavaScript, Java, TypeScript, Bash, SQL, C, C++, PHP, HTML, and more |