ENPL– Polski Check your hardware

Pick a card. See the difference.

Graphics card
Model

GeForce RTX 3090 Qwen3.8-27B UD-Q4_K_XL · 17.6 GB

Typical app 42.9 t/s
Best flags 75.3 t/s
Our engine 93.3 t/s

Conditions: spec_sweep through the server, the same script on both sides: chat, 6 prompts × 3 seeds, T=1, fresh server, median; ABBA order, mean of the two passes best flags and our engine: 200k context (-c 204800), KV q4_0, MTP n=4, -fa on; ours adds the 64k draft vocabulary, lookup W16, GPU verify; A0: plain -m (-fit: 90k context, KV f16); card at 335 W. spec_sweep through the server, the same script on both sides: chat, 6 prompts × 3 seeds, T=1, fresh server, median; ABBA order, mean of the two passes best flags and our engine: 200k context (-c 204800), KV q4_0, MTP n=4, -fa on; ours adds the 64k draft vocabulary, lookup W16, GPU verify; A0: plain -m (-fit: 90k context, KV f16); card at 335 W. /; :59

2.2×

our engine vs: Typical app

A 500-word answer · derived from t/s

Typical app13.6s Our engine6.3s

done

7.4 s faster

The same script, other tasks: greedy chat (T=0) 42.3 → 76.7 → 103.4 t/s; code edits 42.0 → 109.5 → 348.9 t/s; copying 41.8 → 108.4 → 343.8 t/s; generation after a 32k context 39.1 → 64.0 → 79.9 t/s. Honestly: with a 200k context our engine did not fit a ~55.6k-token prompt in 24 GB; on 24 GB NVIDIA cards we set the context to ≤ 131,072 (22.2 GB peak at 60k).

The same answer, at the measured pace

GeForce RTX 3090, Qwen3.8-27B UD-Q4_K_XL: Typical app 42.9 tokens per second, Our engine 93.3 tokens per second.

Lengths: the time for a 500-word answer (≈ 585 tokens) is derived from the pace. The race is an example answer of 59 words (≈ 69 tokens) at the same pace. The pace itself was measured under the conditions given under the bars.

See everything about your card