Qwen3.6 35B (3B active)
on NVIDIA RTX 3060 12GB · Strata · 4,096 ctx
Summary
User benchmarks Qwen3.6-35B-A3B at 71.5 t/s decode and 1539 t/s prefill on an RTX 3060 12GB, comparing Strata against llama.cpp.
Setup is Strata with UD-Q2_K_XL GGUF, 4096-token prompt, 128 generated tokens, MTP off, mean of 2 runs; the 11.7 GB model exceeds the 12 GB card so weights ran partly from system RAM.
llama.cpp on the same file gave 39.6 t/s decode and 1029 t/s prefill. With MTP on, the 35B reaches about 160 t/s decode, and a hand-tuned llama.cpp (-ncmoe 12) on a quiet machine reached about 2236 t/s prefill and 68 t/s decode. The table also covers Qwen3.5-2B, Qwen3-4B, Qwen3.8-27B Swift-1.5, Qwen3-30B-A3B, Qwen3-Next-80B-A3B and Qwen3.8-Flash-Next (177B).
Compare this run
Medians of plain runs: one device, one request, no speculative decoding, the whole model in the device's memory. The report above is one setup, so read it beside them and not as one of them.
Qwen3.6 35B · 3B active at 4-bit by hardware
- NVIDIA RTX 509032 GB215.1 t/s · one run
- 178.0 t/s · one run
- 137.7 t/s · 2 runs
- AMD RX 7900 XTX24 GB125.5 t/s · one run
- NVIDIA RTX 309024 GB101.7 t/s · one run
- 81.1 t/s · one run
Bars are medians on one scale, 0 to 215.1 tokens/s. Each dot is one run and opens its report. Press ▶ to watch text arrive at a row's speed.
Models most run on the NVIDIA RTX 3060 12GB
- Gemma 4 8B4-bit59.5 t/s · 2 runs
- Gemma 4 5.1B4-bit60.0 t/s · one run
- Qwen2.5 7B4-bit49.0 t/s · one run
- 113.6 t/s · one run
Bars are medians on one scale, 0 to 113.6 tokens/s. Each dot is one run and opens its report. Press ▶ to watch text arrive at a row's speed.
Everything reported on the NVIDIA RTX 3060 12GB →Every Qwen3.6 report →Estimate Qwen3.6 on this card →