Measured model
Measured on a Mac Studio (M5 Max, 36GB)Hy-Memory-30B-A3B
Tencent · Ollama name hf.co/tencent/Hy-Memory-30B-A3B-GGUF:Q4_K_M
On a Mac Studio (M5 Max, 36GB), Hy-Memory-30B-A3B answers a simple question in about 0.1 s, gets 66% of the Japanese knowledge questions right, 88% of the math problems, and uses 19.2GB of memory.
Results
| Measure | Result | Among measured models | Note |
|---|---|---|---|
| Reply time | about 0.1 s | 4 of 38 | Until one simple common-sense question is answered (median of 50) |
| Knowledge (high school to university) | 66% | 34 of 38 | 200 questions, ±7 |
| Math (word problems) | 88% | 29 of 38 | 100 questions, ±8 |
| Memory used | 19.2GB | 31 of 38 | |
| EN→JA translation | 31 pts | 18 of 34 | out of 100 (chrF), about 0.5 s a sentence |
| JA→EN translation | 56 pts | 13 of 34 | out of 100 (chrF), about 0.3 s a sentence |
"Among measured models" is the place among the 38 models measured on this site (lower is better for time and memory).
Measured on other Macs
The numbers and places on this page come from the Mac above alone.
Try it on your Mac
- Install the Mac version of Ollama from ollama.com and start it.
- Type this in Terminal. The first time, the model downloads; then you can chat with it.
ollama run hf.co/tencent/Hy-Memory-30B-A3B-GGUF:Q4_K_M
It used 19.2GB here. To use it comfortably alongside other apps, a Mac with 48GB of memory or more is a good guide (the model then takes about half the memory).
About the model
- Maker
- Tencent
- Size
- 30.1B parameters
- Quantization
- Q4_K_M
- Thinks before answering
- Yes
- Longest input at once
- 262,144 tokens
- Released (Hugging Face)
- —
- Generation speed
- 107.4 tok/s
- Start-up (loading)
- about 5.3 s
- Measured on
- Oct 21, 2026
- Ollama
- 0.34.3
- Hugging Face
- tencent/Hy-Memory-30B-A3B
What the parts of a name mean (size, quantization and so on): how to read a model name.
Models of similar memory
- qwen3-coder:30b (19.4GB of memory, knowledge 69%, reply about 0.1 s)
- glm-4.7-flash:latest (19.5GB of memory, knowledge 76%, reply about 7.8 s)
- gemma4:26b (18.7GB of memory, knowledge 85%, reply about 2.9 s)
- qwen3.8:27b (18.2GB of memory, knowledge 87%, reply about 7.7 s)
- gemma3:27b (18GB of memory, knowledge 76%, reply about 0.4 s)
Compare the main models on the text page. The series page lists every model measured.