Apple M4 (10-core GPU, 24GB) × Gemma 3 27B
Can the Apple M4 (10-core GPU, 24GB) run Gemma 3 27B?
Barely
Yes. The Apple M4 (10-core GPU, 24GB) can run Gemma 3 27B at up to Q3_K_M (needs ~16.6 GB at 4K context of 18.0 GB usable), at an estimated ~6.9 tokens/sec at Q3_K_M.
Speed is estimated from memory bandwidth calibrated with a measured 24.11 tok/s Llama 2 7B Q4_0 result (source). No direct public benchmark of this exact pair is recorded yet.
By quantization (8K context)
| Quant | Memory needed | Fits? | Est. tok/s |
|---|---|---|---|
| Q3_K_M | 18.7 GB | ✅ at 4K ctx | ~6.9 |
| Q4_K_M | 22.1 GB | ❌ | — |
| Q5_K_M | 25.1 GB | ❌ | — |
| Q6_K | 28.2 GB | ❌ | — |
| Q8_0 | 34.9 GB | ❌ | — |
| F16 | 61.4 GB | ❌ | — |
Fastest hardware for Gemma 3 27B
- NVIDIA GeForce RTX 5090 — ~66.7 tok/s at Q4_K_M
- NVIDIA GeForce RTX 4090 — ~42.8 tok/s at Q4_K_M
- AMD Radeon RX 7900 XTX — ~42.4 tok/s at Q4_K_M
- NVIDIA GeForce RTX 3090 Ti — ~39.4 tok/s at Q4_K_M
- NVIDIA GeForce RTX 3090 — ~36.4 tok/s at Q4_K_M
No suitable GPU? Rent one by the hour on Vast.ai or RunPod (affiliate links).
Get new models & benchmarks in your inbox
No spam. Unsubscribe anytime by replying to any email.