Inco Splash benchmarks
An inference engine for running open-weight LLMs locally.
16 community reports
Inco Splash results by GPU
Hardware with reported Inco Splash results, with each report's model, quant and speed, newest first. Runs on several cards, with speculative decoding, with batched requests or with part of the model in system RAM say so, since each describes a different setup.
Inco Splash on M1 Max 64GB4 reports
- Qwen3.8 27B · 4-bit group-6433.1 t/s(speculative)
- Qwen3.6 35B · 3B active · 4-bit144.0 t/s(speculative)
- Qwen3.8 27B · 4-bit32.8 t/s, 101 t/s prompt(speculative)
- Qwen3.8 27B · 4-bit39.1 t/s, 142 t/s prompt(speculative)
Inco Splash on M5 Pro 64GB4 reports
- Qwen3.8 27B · oQ4e76.2 t/s, 393 t/s prompt(speculative)
- Qwen3.8 27B · Q8_035.3 t/s, 529 t/s prompt(speculative)
- Qwen3.8 27B · UD-Q4_K_XL50.0 t/s(speculative)
- Qwen3.8 27B · Q836.9 t/s(speculative)
Inco Splash on M5 Max 128GB2 reports
- Qwen3.8 125B · 6B active · Q5_K_S116.6 t/s
- Qwen3.8 27B · 4bit87.8 t/s(speculative)
Inco Splash on M5 Ultra 96GB2 reports
- Qwen3.8 125B · 6B active · UD-IQ4_XS145.6 t/s, 3,880 t/s prompt(speculative)
- Qwen3.8 125B · 6B active · IQ3_S144.6 t/s, 3,444 t/s prompt(speculative)
Inco Splash on M3 Max 128GB1 report
- Qwen3.6 35B · 3B active · 4-bit98.8 t/s, 2,358 t/s prompt(speculative)
Inco Splash on M3 Max 48GB1 report
- Qwen3.8 27B · 4-bit91.8 t/s(speculative)
Inco Splash on M4 Max 36GB1 report
- Qwen3.8 27B15.0 t/s, 100 t/s prompt(speculative)
Inco Splash on M5 24GB1 report
- Qwen3.8 125B · 6B active · Q2_049.0 t/s(speculative, part in system RAM)
Inco Splash on M5 Max 36GB1 report
- Qwen3.8 27B144.0 t/s(speculative)
Inco Splash on M5 Pro 24GB1 report
- Qwen3.8 27B · UD-IQ3_XXS55.0 t/s, 448 t/s prompt(speculative)
Inco Splash, GPU not identified3 reports
- Qwen3.8 27B · 4-bit89.8 t/s(speculative)
- Qwen3.6 35B · 3B active144.0 t/s(speculative)
- Qwen3.8 27B74.0 t/s(speculative)
This engine doesn't yet have an editorial profile on llamaperf. The community reports below show how it's been used in practice across different hardware.
Compare GPUs running Inco Splash
No plain run (one device, one request, no speculative decoding, the whole model in memory) with a generation speed has been reported on Inco Splash yet, so there is no median to compare. The reports behind each card are listed below.
| GPU | VRAM | Reports |
|---|---|---|
| M1 Max 64GBapple | 64GB | 4 |
| M5 Pro 64GBapple | 64GB | 4 |
| M5 Max 128GBapple | 128GB | 2 |
| M5 Ultra 96GBapple | 96GB | 2 |
| M3 Max 128GBapple | 128GB | 1 |
| M3 Max 48GBapple | 48GB | 1 |
| M4 Max 36GBapple | 36GB | 1 |
| M5 24GBapple | 24GB | 1 |
Inco Splash against other engines
Pairs of plain runs on the same card, of the same model size at the same bit class: one device, one request, no speculative decoding, the whole model in memory. Context length and build still differ between the two sides, and each side shows its own.
No matched pair yet. No card has plain runs of one model size at one bit class on Inco Splash and on another engine, so llamaperf can't say how it compares on speed. Add a run.
Top models on Inco Splash
Frequently asked
Is Inco Splash faster than other engines?
llamaperf has no matched comparison for Inco Splash yet: no card has plain runs of the same model size at the same bit class on Inco Splash and on another engine. Speed claims about engines need that pairing, so this page doesn't make one.