Leaderboard / Tencent (Hunyuan)

Hy-Memory 30B-A3B

Tencent (Hunyuan) · Hy-Memory · released 2026-10-15 · data checked 2026-11-13

Hy-Memory 30B-A3B is a 30.1B-parameter Mixture-of-Experts model (3B active per token) from Tencent (Hunyuan) (China), released 2026-10-15 under the Apache-2.0 license. It doesn't have enough reported benchmarks for an index score yet. It needs about 20.2 GB of memory at 4-bit with an 8K-token context.

Long-term-memory writer: replaces the LLM inside memory frameworks (Mem0, MemOS, Hy-Memory) to decide what to store and how to update it. Not a general assistant; it reports only memory benchmarks (LoCoMo, LongMemEval, PersonaMem-v2), where it matches Claude Opus 5 as a writer.

Small LM Index
—
Needs 2 core benchmarks (has 0)
Parameters
30.1B
3B active · MoE
Price per 1M tokens
—
No hosted API on OpenRouter; self-host
Context window
262K
tokens

Benchmarks

Each bar shows this model's score. The tick marks the best medium model on that benchmark.

No benchmark results for this release yet. Know of one? Send it.

Specifications

Lab
Tencent (Hunyuan) (CN · China)
Architecture
Mixture-of-Experts
Total parameters
30.1B
Active parameters
3B
Size class
Medium
Context window
262K tokens
Input modalities
text
Reasoning
Hybrid (thinking on/off)
License
Apache-2.0
Openness
Open weights
Memory to run
~20.2 GB at 4-bit, 8K context (more sizes)
Hugging Face
6.6K downloads (30 days) · 326 likes

Compare with

Sources

Something wrong here? Report it.

Run it locally

Uses tencent/Hy-Memory-30B-A3B-GGUF (Q4_K_M). Swap the tag for another size: Q3_K_M, Q4_K_M, Q6_K, Q8_0.

Ollama
ollama run hf.co/tencent/Hy-Memory-30B-A3B-GGUF:Q4_K_M
llama.cpp
llama-server -hf tencent/Hy-Memory-30B-A3B-GGUF:Q4_K_M
Then open http://localhost:8080 for a chat window.
MLX (Mac)
pip install -U mlx-lm
mlx_lm.chat --model mlx-community/Hy-Memory-30B-A3B-4bit
vLLM (GPU server)
vllm serve tencent/Hy-Memory-30B-A3B
Full-precision weights, for a server GPU.

Open in LM Studio

Memory needed: weights, context cache and runtime buffers.

Quantization4K context32K context256K context
16-bit F1662 GB65 GB87 GB
8-bit Q8_033 GB36 GB59 GB
4-bit Q4_K_M20 GB23 GB45 GB

Check it against your GPU, Mac or phone →