Core Development: Bee 2.6 and items for Bee 2.7
Bee 2.6 has delivered a more stable and responsive network, with fewer freezes, faster upload speeds, and a more even distribution of nodes.
Next up, the Bee team is working on version 2.7. The release will include a WebAssembly implementation for browser-native nodes with automatic TLS handling, support for multiple underlay addresses, and the introduction of secure node certificates.
Development is also progressing on new metrics for pull-sync analysis, alongside research into protocol improvements such as negative incentives and smart neighbourhood management. Each release continues to be metrics-driven, ensuring that changes bring measurable improvements to stability and performance.
In focus: The path ahead
Accessibility and Usability
Improvements in Bee 2.6 have cut node sync times by up to 80% in testing, significantly lowering the barrier for new operators. Postage stamp syncing has also been improved, with further refinements planned.
Meanwhile, work continues on the new Swarm Desktop file manager, which will integrate ACT (access control trie) to simplify local file syncing. Another major feature in progress is global pinning, which will allow content pinned locally to be directly served by the network.
This effectively reintroduces the model of self-hosting—offering a practical, no-cost entry point for applications, while still allowing users to scale into paid storage when needed.
Ecosystem Services: Swarmy, Beeport, and MCP
Updates were also shared on ecosystem services:
- Swarmy continues to develop as a service provider for users who prefer not to operate their own nodes. \
- Beeport (formerly known as the multichain app) is in testing, with a focus on scaling, liquidity, and integration. Once live, it will allow users to purchase storage directly from any chain, and it will be integrated into the new Swarm Desktop.
- The Solar Punk (ecosystem partner) team presented progress on the Model Context Protocol (MCP), which is being developed as a decentralised data streaming layer for AI and multimedia. Following a successful proof of concept, the team is now working toward an MVP. Future plans for MCP include decentralised agent discovery, private data sharing via ACT, and integration with existing AI frameworks.