Llama 3.2 1B Instruct › Apple M4 Pro (20-core GPU)
Llama 3.2 1B Instruct on the Apple M4 Pro (20-core GPU)
Best measured: 119 tokens/s generation and 2,131 tokens/s prompt processing (Q4_K_M, llamafile).
For comparison, our size-scaling estimate for this combination is 193 tok/s.
Every run
| Quant | Backend | Generate tok/s | Prompt tok/s | Test | Build / by / notes | Source |
|---|---|---|---|---|---|---|
| Q4_K_M | llamafile | 120 | 2,131 | LocalScore avg | LocalScore rank 148/878 | LocalScore (localscore.ai) |
| Q4_K_M | llamafile | 119 | 2,145 | LocalScore avg | LocalScore rank 149/878 | LocalScore (localscore.ai) |
| Q4_K_M | llamafile | 112 | 2,037 | LocalScore avg | LocalScore rank 153/878 | LocalScore (localscore.ai) |
tg128 = generating 128 tokens; pp512 = reading a 512-token prompt (llama-bench). LocalScore averages several prompt and output lengths, so its numbers run lower than tg128.
No card yet?
Try it on a rented GPU (RunPod)Affiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.
All results for the Apple M4 Pro (20-core GPU) · Llama 3.2 1B Instruct on other hardware