Benchmark result

qwen3.8-flash-next-iq3_s on 2× NVIDIA GeForce RTX 3080 — 120.3 tok/s

Measured with Strata Strata 0.1.52 on October 30, 2026.

ShareRedditX

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

Reproduce this run

toolStrata Strata 0.1.52
modelqwen3.8-flash-next-iq3_s
context256,000
thinkingon
effortmedium
samplingtemperature 1 · top_p 0.95 · top_k 20
clientv0.4.98+102

Set those in Strata, then:

llm-benchmark benchmark --model "qwen3.8-flash-next-iq3_s" --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.