Llama 3.1 8B Instruct › Apple M2 Max (38-core GPU)
Llama 3.1 8B Instruct on the Apple M2 Max (38-core GPU)
Best measured: 45.5 tokens/s generation and 479 tokens/s prompt processing (Q4_K_M, llamafile).
For comparison, our size-scaling estimate for this combination is 67.5 tok/s.
Every run
| Quant | Backend | Generate tok/s | Prompt tok/s | Test | Build / by / notes | Source |
|---|---|---|---|---|---|---|
| Q4_K_M | llamafile | 46.4 | 484 | LocalScore avg | LocalScore rank 110/422 | LocalScore (localscore.ai) |
| Q4_K_M | llamafile | 44.7 | 474 | LocalScore avg | LocalScore rank 112/422 | 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 M2 Max (38-core GPU) · Llama 3.1 8B Instruct on other hardware