Benchmark result
qwen3.8-flash-next-unsloth-ud-q4_k_xl on NVIDIA GeForce RTX 3090 — 55.8 tok/s
Measured with Strata Strata 0.1.60 on November 18, 2026.
Support llm-bench.io
llm-bench.io is free. If it helped you, I'd appreciate your support to cover the infrastructure costs.
What the model built
Open full screen →The coding scenario asks for a playable game in a single HTML file. This is exactly what the model returned, unedited; the frame only adds a line that tells this page how tall it is.
How this run compares
6th fastest of 20 runs of this model on NVIDIA GeForce RTX 3090 · median 52.4 tok/s.
sort
- 156.9tok/s84.1Strata · Q4_K
- 256.9tok/s88.7Strata · Q4_K
- 356.8tok/s79.8Strata · Q4_K
- 456.5tok/s86.9Strata · Q4_K
- 555.9tok/s87.2Strata · Q4_K
- 655.8tok/s86.9Strata · Q4_Kthis run
- 754.7tok/s83.5Strata · Q4_K
- 854.6tok/s82.8Strata · Q4_K
- 954.5tok/s88.9Strata · Q4_K
- 1052.8tok/s84.9Strata · Q4_K
- 1152.0tok/s83.7Strata · Q4_K
- 1251.0tok/s85.2Strata · Q4_K
- 1350.2tok/s84.0Strata · Q4_K
- 1448.0tok/s89.3Strata · Q4_K
- 1547.8tok/s85.2Strata · Q4_K
- 1647.7tok/s86.1Strata · Q4_K
- 1726.8tok/s82.9Strata · Q4_K
- 1825.0tok/s79.4Strata · Q4_K
- 1924.8tok/s81.6Strata · Q4_K
- 2021.3tok/s85.5Strata · Q4_K
qwen3.8-flash-next-unsloth-ud-q4_k_xl on other hardware
Reproduce this run
toolStrata Strata 0.1.60
modelqwen3.8-flash-next-unsloth-ud-q4_k_xl (Q4_K)
context160,000
thinkingon
effortxhigh
samplingtemperature 1 · top_p 0.95 · top_k 20 · min_p 0 · presence_penalty 0 · repeat_penalty 1
clientv0.4.98+102
Set those in Strata, then:
llm-benchmark benchmark --model "qwen3.8-flash-next-unsloth-ud-q4_k_xl" --tool "Strata"The client prompts for context length and thinking mode, and for the KV cache dtype on Unsloth Studio. Sampling is left at the model default — the values above are what the tool reported using, not overrides the client sent.