Hy-Memory-30B-A3B
- Open weights
Prices
0 offerings · prices as listedNo priced offering is listed for this model yet.
Details
- Licence
- apache-2.0
- Family
- hy-memory
Not stated on Tencent’s pages: max input, input modalities, output modalities, reasoning, reasoning efforts, reasoning can be off, max thinking budget, tool use, structured output, built-in tools, knowledge cutoff, aliases.
Run locally
Weights: tencent/Hy-Memory-30B-A3B- Parameters 30.1B
- Active 3B · 128 experts, 8 per token
- Native precision BF16
- Layers 48
- KV cache per token 96 KB
- Config positions 256K (config.json max_position_embeddings)
- API context 63K (maker’s API)
- Needs
- ≈ 18.6 GB (weights 17.0 (maker’s file) + KV cache 0.8 + overhead 0.8)
- Usable
- 21.6 GB (90% of its memory)
- Generation speed
- 192–288 tok/s (reads the active 3B per token)
- Longest context
- 40K (longer doesn’t fit)
- First token
- about 0.10–0.15 s for a 2K prompt
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| BF16 (native)maker’s file | 56.0 GB | ≈ 59.6 | ○Needs system memory 12–18 tok/s | — |
| Q8_0maker’s file | 29.8 GB | ≈ 32.0 | ○Needs system memory 34–51 tok/s | — |
| FP8maker’s file | 28.5 GB | ≈ 30.7 | ○Needs system memory 37–56 tok/s | — |
| Q6_Kmaker’s file | 23.0 GB | ≈ 24.9 | ○Needs system memory 82–123 tok/s | — |
| Q5_K_M | 19.9 GB | ≈ 21.7 | ○Needs system memory 195–292 tok/s | — |
| Q5_0 | 19.2 GB | ≈ 21.0 | ●Runs well 176–264 tok/s | 14K |
| Q4_K_Mmaker’s file | 17.0 GB | ≈ 18.6 | ●Runs well 192–288 tok/s | 40K |
| Q4_0 | 15.7 GB | ≈ 17.3 | ●Runs well 202–303 tok/s | 54K |
| IQ4_XS | 14.9 GB | ≈ 16.4 | ●Runs well 210–315 tok/s | 63K |
| Q3_K_Mmaker’s filequality drops | 13.5 GB | ≈ 14.9 | ●Runs well 224–336 tok/s | 63K |
| IQ3_XXSquality drops | 10.7 GB | ≈ 12.0 | ●Runs well 258–387 tok/s | 63K |
| Q2_Kquality drops | 10.5 GB | ≈ 11.8 | ●Runs well 261–392 tok/s | 63K |
| IQ2_XXSquality drops | 7.2 GB | ≈ 8.5 | ●Runs well 319–479 tok/s | 63K |
Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.
Compared with other hardware
Each at the highest precision it runs. Click one for details.
Smallest that runs it (Q4_K_M, 8K): GPU GeForce RTX 4090; unified memory Apple M6 (24GB or 32GB) · 32 GB.
Run commands
llama-server -hf tencent/Hy-Memory-30B-A3B-GGUF:Q4_K_M -c 8192
ollama run hf.co/tencent/Hy-Memory-30B-A3B-GGUF:Q4_K_M
Memory to fine-tune
Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.
Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-06. How it’s estimated
Cited scores
From third-party leaderboards, not measured by usThis model isn’t on any leaderboard we cite yet.
Changes
All changes →- New language model
Sources
- tencent/Hy-Memory-30B-A3B config.json (opens in a new tab) huggingface.co · checked Nov 2, 2026
- tencent/Hy-Memory-30B-A3B model card (opens in a new tab) huggingface.co · checked Nov 3, 2026
- tencent/Hy-Memory-30B-A3B model card (opens in a new tab) huggingface.co · checked Nov 3, 2026
- tencent/Hy-Memory-30B-A3B README.md (opens in a new tab) huggingface.co · checked Oct 31, 2026
- tencent/Hy-Memory-30B-A3B-FP8 files (opens in a new tab) huggingface.co · checked Nov 2, 2026
- Hy-Memory-30B-A3B-GGUF (opens in a new tab) huggingface.co · checked Nov 5, 2026
- tencent/Hy-Memory-30B-A3B files (opens in a new tab) huggingface.co · checked Nov 4, 2026
- tencent/Hy-Memory-30B-A3B-GGUF files (opens in a new tab) huggingface.co · checked Nov 2, 2026
Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data
Tencent · China · tencent.com
Related models
Snapshots, the same family and other models from this provider