How fast is swift-1.5-qwen3.8-flash-next-iq3_s for local inference?

Across 2 community benchmark runs, swift-1.5-qwen3.8-flash-next-iq3_s reaches up to 310.6 tok/s and averages 259.0 tok/s, with the fastest results on NVIDIA GeForce RTX 5090.

Which tools have been used to run swift-1.5-qwen3.8-flash-next-iq3_s?

Benchmarks were submitted using Strata, basalt. Results are community-contributed and updated as new runs arrive.

Model benchmarks

swift-1.5-qwen3.8-flash-next-iq3_s local LLM performance

As of October 2026, swift-1.5-qwen3.8-flash-next-iq3_s runs at up to 310.6 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

Stratabasalt

ShareReddit X

Model size

Unknown

Peak speed

310.6 tok/s

Average speed

259.0 tok/s

Avg PP

1573.4 tok/s

Min memory

n/a

Max context

65,536 tokens

Avg output / run

51,600 tokens

Avg runtime / run

3m 10s

Avg quality

86.4

Benchmark runs

GPUs tested

Quality by task

Average LLM-judged quality (0–100) with the run-to-run spread shown as a P5–P95 band, overall and for each benchmark task, across all 2 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.

Task	P5 (low)	Avg	P95 (high)
Overall	86.3	86.4	86.4
Agent Workflow	87.0	88.9	90.9
Code Generation	68.7	74.6	80.4
Role Play & Narrative	91.6	92.5	93.4
Research & Analysis	86.7	89.4	92.1

Performance by hardware and tool

Every hardware/tool/quantization combination swift-1.5-qwen3.8-flash-next-iq3_s has been benchmarked on, ranked by peak token generation speed. Last updated October 2026.

Hardware	Tool	Quant	Best tok/s	Avg tok/s	Memory	Context	Quality	Runs
NVIDIA GeForce RTX 5090	basalt	—	310.6 tok/s	310.6 tok/s	n/a	65,536 tokens	86.4	1
NVIDIA GeForce RTX 5090	Strata	—	207.5 tok/s	207.5 tok/s	n/a	65,536 tokens	86.3	1

Benchmark runs

All 2 swift-1.5-qwen3.8-flash-next-iq3_s runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

How fast is swift-1.5-qwen3.8-flash-next-iq3_s for local inference?: Across 2 community benchmark runs, swift-1.5-qwen3.8-flash-next-iq3_s reaches up to 310.6 tok/s and averages 259.0 tok/s, with the fastest results on NVIDIA GeForce RTX 5090.
Which tools have been used to run swift-1.5-qwen3.8-flash-next-iq3_s?: Benchmarks were submitted using Strata, basalt. Results are community-contributed and updated as new runs arrive.