Medusa has added a migration tool to its Medusa MCP server that lets users prompt an AI agent to migrate store data from Shopify directly into Medusa, covering products, variants, prices, images, inve…
Medusa, the open-source commerce platform, has added two mechanisms to its MCP server to capture feedback from coding agents: an explicit feedback tool that reports bugs and documentation gaps to Post…
Speculative decoding can speed up local large language model inference by 1.5 to 2.5 times without changing output quality, according to research from Google and DeepMind. The technique uses a small d…
Medusa v2.18 was released on July 23, 2026, with new features, developer experience improvements, and bug fixes, including community contributions that doubled unique monthly contributors from about 1…
Oli Juhl of Medusa recommends that every Model Context Protocol (MCP) author add two key features to their MCP servers, based on Medusa's experience providing coding agents with documentation access. …
Researchers propose EntMTP, a training-free scheduler that dynamically adjusts multi-token prediction depth based on local entropy, achieving up to 1.36x speedup over Medusa baselines in LLM inference…
Researchers introduced JetSpec, a speculative decoding method that trains a causal parallel draft head on a frozen target model to draft entire speculative trees in one pass while preserving autoregre…