Compare
Loading community benchmarks…
Loading community benchmarks…
Put two chips, runtimes, or runtime versions side by side. Everything else is held fixed or called out, so a ratio is worth exactly what the matched runs say.
Ratios read side A relative to side B. Geometric mean over 1 like-for-like pair.
Everything else matches: model, quantisation, backend, protocol, and workload are the same on both sides.
Each row is a configuration present on both sides. Values are per-cell medians; ratios read Apple M1 Max relative to Apple M3 Ultra.
| Configuration | Decode A | Decode B | Ratio | Prefill A | Prefill B | Ratio | Runs A / B |
|---|---|---|---|---|---|---|---|
Mistral-7B-Instruct-v0.3 BaseRTQ4 | 53.5 | 104.1 | 0.51× | 510 | 1,414 | 0.36× | 2 / 2 |
Top 2 of 2 runs by decode throughput.
| Model / format | Chip / backend | Decode | Prefill | Contributor | Date | Report |
|---|---|---|---|---|---|---|
Mistral-7B-Instruct-v0.3-Q4 BaseRT 0.2.6Q4 | Apple M1 Max Metal | 66.4 TG128 @ 1 ctx | 511 PP512 | basecompute | View Benchmark | |
Mistral-7B-Instruct-v0.3-Q8 BaseRT 0.2.6Q4 | Apple M1 Max Metal | 40.6 TG128 @ 1 ctx | 509 PP512 | basecompute | View Benchmark |
Top 2 of 2 runs by decode throughput.
| Model / format | Chip / backend | Decode | Prefill | Contributor | Date | Report |
|---|---|---|---|---|---|---|
Mistral-7B-Instruct-v0.3-Q4 BaseRT 0.2.6Q4 | Apple M3 Ultra Metal | 129.3 TG128 @ 1 ctx | 1,440 PP512 | basecompute | View Benchmark | |
Mistral-7B-Instruct-v0.3-Q8 BaseRT 0.2.6Q4 | Apple M3 Ultra Metal | 78.9 TG128 @ 1 ctx | 1,388 PP512 | basecompute | View Benchmark |