30B-A3B Tencent Hy
Tencent's memory-writing model for agents, released in October under Apache 2.0: it is built to turn an agent's history into stored memories, not to chat. It is a 30B mixture-of-experts model with 128 experts, 8 used per token, so about 3B parameters are read for each token. Every layer keeps a full cache, 96 KB per token. Tencent publishes its own GGUF files, including two mixed builds that keep everything at Q8_0 except the experts, which are stored at IQ2_XXS or IQ1_M. No quality loss has been published for any build.
262K
Max Context
6
Quant Variants
GGUF Q8_0
Best Quality
—
Perplexity rise (best build)
Quantization Variants #
Per-quant VRAM, quality loss, and inference speed on RTX 4090
Measured = run on this site · Estimated = not run here: calculated, or a figure this site has not verified · Community = public reports
| Format | Level | BPW | VRAM | PPL Loss | Speed | Source | Actions |
|---|---|---|---|---|---|---|---|
| GGUF | Q3_K_M | 3.84 | 15.5 GB | — | — | Estimated | |
| GGUF | Q4_K_M | 4.85 | 19.5 GB | — | — | Estimated | |
| GGUF | Q6_K | 6.57 | 26.2 GB | — | — | Estimated | |
| GGUF | Q8_0 | 8.51 | 33.8 GB | — | — | Estimated | |
| GGUF | Q8_0-expsIQ2_XXS | 2.43 | 9.9 GB | — | — | Estimated | |
| GGUF | Q8_0-expsIQ1_M | 2.14 | 8.8 GB | — | — | Estimated |
Which file each size is for
A file size is not a memory figure: the VRAM column is the calculator's total at 4K context — weights plus KV cache plus a 10% activation buffer. Units differ on purpose: file sizes are decimal GB, as Hugging Face lists them; the VRAM column is GiB (1024³ bytes) with 2% added to the weights, so 7.12 GB of file counts as about 6.8 GiB there. A vision projector (mmproj), where the repo ships one, is a separate download and is in neither number.
- GGUF Q3_K_M[tencent/Hy-Memory-30B-A3B-GGUF](https://huggingface.co/tencent/Hy-Memory-30B-A3B-GGUF) · Hy-Memory-30B-A3B-Q3_K_M.gguf · 14.45 GB (14,446,329,600 bytes) · listing read 2026-11-12
- GGUF Q4_K_M[tencent/Hy-Memory-30B-A3B-GGUF](https://huggingface.co/tencent/Hy-Memory-30B-A3B-GGUF) · Hy-Memory-30B-A3B-Q4_K_M.gguf · 18.24 GB (18,236,702,464 bytes) · listing read 2026-11-12
- GGUF Q6_K[tencent/Hy-Memory-30B-A3B-GGUF](https://huggingface.co/tencent/Hy-Memory-30B-A3B-GGUF) · Hy-Memory-30B-A3B-Q6_K.gguf · 24.71 GB (24,707,464,960 bytes) · listing read 2026-11-12
- GGUF Q8_0[tencent/Hy-Memory-30B-A3B-GGUF](https://huggingface.co/tencent/Hy-Memory-30B-A3B-GGUF) · Hy-Memory-30B-A3B-Q8_0.gguf · 31.99 GB (31,985,728,896 bytes) · listing read 2026-11-12
- GGUF Q8_0-expsIQ2_XXStencent/Hy-Memory-30B-A3B-GGUF · Hy-Memory-30B-A3B-Q8_0-expsIQ2_XXS.gguf · 9.14 GB (9,143,025,408 bytes) · listing read 2026-11-12
- GGUF Q8_0-expsIQ1_Mtencent/Hy-Memory-30B-A3B-GGUF · Hy-Memory-30B-A3B-Q8_0-expsIQ1_M.gguf · 8.03 GB (8,034,156,288 bytes) · listing read 2026-11-12
Similar models #
[Compare with Qwen3-VL 30B-A3B](https://quantized.uk/tools/compare/?a=hy-memory-30b-a3b&b=qwen3-vl-30b-a3b)
[30B-A3BConsumer GPUPro GPUQ4_K_M19.9 GB+2.6%perplexity rise](https://quantized.uk/quant-hub/qwen3-vl-30b-a3b/)
### Qwen3-VL 30B-A3B Instruct
Alibaba Qwen3-VL
Multimodal MoE with only ~3B active parameters, so it stays responsive on Apple unified memory and survives CPU offload far better than a dense 30B. At Q4_K_M it needs about 19.9 GB at 4K, comfortable on a 24 GB card; Q3_K_M brings that to 15.1 GB. The vision encoder is a separate 1.1 GB file.
36BConsumer GPUPro GPUQ4_K_M23.9 GB+2.7%perplexity rise
Seed-OSS 36B Instruct
ByteDance Seed
Dense 36B with a native 512K context. The Q4_K_M file is 21.8 GB, so with cache and buffers it needs about 23.9 GB at 4K: tight on a 24 GB card, where Q3_K_M (19.5 GB) is the comfortable choice, or a 2×16GB split. The AWQ build is 21.1 GB, not the 4-bit rate the name suggests. Context is the real cost: 512K of KV cache is roughly 128GB on its own, so budget context first and weights second.
27BConsumer GPUPro GPUQ4_K_M18.5 GB—perplexity rise
Qwen3.8 27B
Alibaba Qwen3.8
Hybrid attention: only 16 of its 64 layers keep a KV cache, which is why a 27B model holds a 262K context in about 16GB of cache rather than 64GB. Sizing it as a conventional stack overstates the cache fourfold — the estimates here count the 16 and match published measurements at 8K, 32K and 262K. Quality loss per level has not been published.
[30B-A3BConsumer GPUPro GPUQ4_K_M19.5 GB—perplexity rise](https://quantized.uk/quant-hub/glm-4.7-flash/)
### GLM-4.7-Flash
Zhipu GLM-4.7
Zhipu's lightweight member of GLM-4.7: 30B total, about 3B active per token, MIT licence. Its attention is MLA, which caches one compressed 576-value vector per token per layer instead of full keys and values — so at 32K context the cache is under 2 GB where a conventional 47-layer model would need several times that. llama.cpp runs it as a DeepSeek-2-style model. Sizes below come from bartowski's GGUF files; quality loss per level has not been published.
Running Hy-Memory-30B-A3B locally #
At Q4_K_M and 4K of context, Hy-Memory-30B-A3B needs about 19.5 GB — 17.3 GB of weights, 0.38 GB of KV cache and a 1.8 GB activation buffer. The smallest cards in this index that clear that comfortably are 24 GB ones — 4 of them, from the RTX 3090 up — and 31 of the 78 cards here do. These are calculated figures, not measurements: the estimate stops counting a card as comfortable at 88% of its VRAM, which is roughly the room a desktop session needs.
What longer context costs #
Going from 4K to 32K adds about 2.62 GB, taking the total to 22.3 GB. Weights do not move with context — only the KV cache does, and it grows linearly, so this is the number to watch when planning for long documents. The model's native window is 256K; holding all of it at Q4_K_M would need about 45 GB.
Which build to download #
This index only tracks Hy-Memory-30B-A3B in GGUF, across 6 levels (Q3_K_M, Q4_K_M, Q6_K, Q8_0, Q8_0-expsIQ2_XXS, Q8_0-expsIQ1_M). No per-level perplexity sweep has been published for this model, so the quality column is empty rather than estimated — within one model, more bits per weight is the only ordering the data supports. No throughput has been measured for this model on this site's hardware. GGUF runs on llama.cpp and Ollama across NVIDIA, AMD and Apple silicon; AWQ and GPTQ target vLLM on CUDA and ROCm; EXL2 is ExLlamaV2 — CUDA only, and now archived, so local use rather than serving.
Common questions #
- How much VRAM does Hy-Memory-30B-A3B need?
- About 19.5 GB at Q4_K_M with 4K of context and batch 1, rising to roughly 22.3 GB at 32K. That figure is weights plus KV cache plus a 10% activation buffer, calculated from the model's architecture rather than measured on a card.
- Will Hy-Memory-30B-A3B run on a 24GB GPU?
- Yes — at Q4_K_M and 4K context it needs about 19.5 GB, which leaves 4.5 GB spare on a RTX 3090, or on any of the other 24 GB cards here (4 in all). That is the smallest size in this index that clears it comfortably; 31 of 78 do.
- Which quantization of Hy-Memory-30B-A3B should I use?
- No published quality comparison exists for this model, so pick by footprint: Q4_K_M is the level most people run, and a higher bits-per-weight level is more faithful. All 6 levels in the index are Q3_K_M, Q4_K_M, Q6_K, Q8_0, Q8_0-expsIQ2_XXS, Q8_0-expsIQ1_M. Note this is a mixture-of-experts model — all parameters must be resident even though only a fraction are active per token, so the memory cost follows the total, not the active count.
Where this model fits #
Sized at Q4_K_M with a 4K context window, smallest card first. Comfortable means the estimate uses at most 88% of the memory.
- Runs comfortably on
- RTX 309024GB · +4.5Radeon RX 7900 XTX24GB · +4.5RTX 409024GB · +4.5Tesla P40 24G24GB · +4.5Mac M5 32G32GB · +4.5
- Tight but possible
- Radeon RX 7900 XT20GB
- Ships in
- GGUF
- A top pick for
- Best local LLM for 12GB
- Try it yourself
- Size it in the calculatorGenerate the run command
How to actually run this #
Deployment guides for this model and this class of hardware.
This page's figures change when the model or the runtime does.Last updated 2026-11-12 RSS → /feed.xml