Radar / AI infrastructure / Tencent Hy-Memory-30B-A3B
Tencent Hy-Memory-30B-A3B: a dedicated open model that decides what agents remember
Tencent Hunyuan open-sourced Hy-Memory-30B-A3B (30B total, ~3B active parameters), a specialist model that replaces general-purpose LLMs as the memory writer inside Mem0, MemOS, and Hy-Memory agent frameworks, under Apache 2.0.
Why it matters
The first purpose-built open memory-writer model — trained with memory on-policy distillation and attributed-memory reinforcement learning — averaging 55.49% across four memory frameworks versus 54.62% for Claude Opus 5 (Tencent data).
What you could build with it
An indie developer building a persistent coding assistant could swap the general-purpose model currently writing memories in a Mem0 or MemOS pipeline for this 3B-active specialist, cutting memory-write costs while improving how the assistant recalls project decisions across sessions.
Does it hold up?
Early but substantive: an independent developer measured the Hy-Memory 2.0 framework on LongMemEval and showed the memory writer is the accuracy bottleneck, which is exactly the gap this specialist model claims to close — but independent scores for the 30B-A3B model itself have not appeared yet.
Built with Tencent Hy-Memory-30B-A3B
- Independent LongMemEval measurement of Hy-Memory (dev.to)dev.to · A developer independently benchmarked Hy-Memory 2.0 on LongMemEval, finding the writer model is the bottleneck: a cheap writer scored 43% while a strong writer reached parity with the no-LLM baseline.
- Tencent Open Sources Hy Memory 2.0 to Give AI Agents Shared Memory (LapaasVoice)lapaasvoice.com · Coverage of the Hy Memory 2.0 release including the dedicated 30B-A3B memory model, 72.6% LoCoMo score claim, and Apache 2.0 licensing.
- Tencent Hunyuan Hy-Memory-30B-A3B release entry (AIGC News timeline)aigc.news · Curated timeline entry confirming the Oct 17 release of the ~3B-active long-term-memory writer model on Hugging Face under Apache 2.0.
Learn more
First spotted on hf: source.
More AI infrastructure
DeepGEMM: DeepSeek's open GPU BLAS kernel libraryDeepSeek's DeepGEMM, a clean and efficient open-source BLAS kernel library for NVIDIA GPUs (up to 1550 TFLOPS on…infra · JEV 0.77OpenViking — self-evolving context database for AI agentsByteDance Volcengine's open-source (AGPL-3.0) context database that unifies agent memory, knowledge RAG, and skills…infra · JEV 0.76Modular open-sources MAX inference server, Mojo stdlib and accelerator kernelsThe core of Modular's unified AI deployment stack is now open: the MAX inference server (OpenAI-compatible endpoints)…infra · JEV 0.76ai-memory - git-backed long-term memory shared across coding agentsRust server and CLI that captures agent sessions via lifecycle hooks, compiles them into a git-backed markdown wiki…infra · JEV 0.76OmniRouteFree MIT AI gateway: one endpoint, 290+ providers and 500+ models with quota-aware auto-fallback.infra · JEV 0.75Context Mode: context-window optimization for AI coding agentsMCP server for AI coding agents that sandboxes verbose tool output and persists session memory, cutting context use by…infra · JEV 0.74
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.