Builds / Apple M4 Pro (20-core GPU, unified)
MacBook Pro M4 Pro 48 GB: Flint 40B-A2.5B 4-bit at 127.9 tok/s (Kapitsa), 126.6 (rec), 79.6 (llama.cpp)
Source: github.com ↗📏 How measured: reported by the original source; RigBase re-checked the numbers at that source
- GPU(s)
- Apple M4 Pro (20-core GPU, unified)
- VRAM / unified memory (GB)
- 48
- CPU
- Apple M4 Pro
- Motherboard
- not stated
- RAM
- 48 GB unified
- PSU
- not stated
- Other
- MacBook Pro 16 (2024, Mac16,7), macOS 27.0.1, AC power, High Power mode; 32K context, 1 slot, T=1.0
- Reported total cost
- not stated
- Power
- not stated
Measured speed (generation tokens/sec)
| Model | Quant | Engine | tok/s | Details |
|---|---|---|---|---|
| Flint 40B-A2.5B | int4 MLX (RecursiveSI/flint-int4) | Kapitsa (PR #286) | 127.9 | |
| Flint 40B-A2.5B | int4 MLX (same files) | rec 0.1.0 (Recursive) | 126.56 | |
| Flint 40B-A2.5B | Q4_1 GGUF (RecursiveSI/flint-int4-gguf) | llama.cpp b11804, all layers on GPU, -fa on | 79.57 |
Kapitsa bench results, 22 Oct 2026. Decode median of 5 runs of 512 tokens. Peak memory: Kapitsa 23.3 GB, llama.cpp 26.9, rec 27.9. At 128K context decode falls to 45.7 (Kapitsa), 35.2 (rec), 23.8 (llama.cpp). Some runs flagged for browser GPU contention (2-12%).
Source: https://github.com/nace-ai/kapitsa/blob/main/bench/results/2026-10-22-flint-m4pro/README.md
Blank fields = not stated in the source. We never estimate build data; GPU street prices above are separately sourced and dated.
Rent before you buy: try this model on a cloud GPU for a few dollars first: Runpod (referral link; sign up with Google to get the new-user bonus)
Quick answers: Can Apple M4 Pro run Flint 40B-A2.5B?