Explore practical recipes for deploying LMCache across different model architectures, serving engines, storage backends, and environments, along with roadmap updates and contribution guidelines for the open-source community.
Recipes are practical deployment guides that show how to launch LMCache in a specific setup, including supported serving engines, compatible LMCache functionalities, and any known limitations or configuration notes.
A mixture-of-experts architecture designed to improve scaling efficiency by activating only a subset of model parameters per token.
A large-scale model architecture built for long-context, agentic, and high-throughput inference workloads.
Google’s open model architecture for efficient instruction-following and multimodal-oriented workloads.
Mistral-family models for general reasoning, coding, and agentic development workflows.
Open-weight GPT-style models for instruction-following and general-purpose inference.
Follow LMCache’s quarterly roadmap to see current priorities, planned improvements, and upcoming development milestones.
Follow LMCache’s quarterly roadmap to see current priorities, planned improvements, and upcoming development milestones.
Whether you’re fixing a bug, improving docs, adding model support, writing tests, or helping other users, there are many ways to contribute to LMCache.
Learn how to open issues, submit pull requests, follow the review process, and contribute code, documentation, tests, or new model support.
New to LMCache? Start with good first issues, documentation improvements, small bug fixes, or community support tasks.
LMCache integrates with the leading inference engines, storage backends, and orchestration layers in the AI ecosystem.
Learn how to open issues, submit pull requests, follow the review process, and contribute code, documentation, tests, or new model support.
Visualize how memory requirements scale with context length and model size to identify potential VRAM bottlenecks and cache offloading opportunities.
Observability suite is currently in development.
Track community-contributed performance benchmarks across diverse hardware configurations and inference engines.
For detailed version matrices, configuration options, and known limitations, refer to the LMCache Documentation.
The latest benchmarks, release notes, and technical deep-dives from the LMCache team and contributors.
Slack, GitHub, Office Hours
