Model benchmarks
ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 local LLM performance
As of November 2026, ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 runs at up to 159.2 tok/s for local inference (best of 3 community benchmark runs across 2 GPUs).
Model size
177B
Peak speed
159.2 tok/s
Average speed
139.6 tok/s
Avg PP
2477.7 tok/s
Min memory
68.8 GB
Max context
65,536 tokens
Avg output / run
31,252 tokens
Avg runtime / run
3m 58s
Avg quality
72.1
Benchmark runs
3
GPUs tested
2
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 3 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 | 62.2 | 72.1 | 82.1 |
| Agent Workflow | 44.6 | 65.2 | 84.5 |
| Code Generation | 41.9 | 66.9 | 83.7 |
| Role Play & Narrative | 64.9 | 77.8 | 86.8 |
| Research & Analysis | 74.6 | 78.5 | 83.9 |
Performance by hardware and tool
Every hardware/tool/quantization combination ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| Apple M5 Ultra | splash | Q2_0 | 159.2 tok/s | 159.2 tok/s | 68.8 GB | 16,384 tokens | 61.1 | 1 |
| Apple M5 Max | splash | Q2_0 | 136.1 tok/s | 129.7 tok/s | 79.2 GB | 65,536 tokens | 77.6 | 2 |
Benchmark runs
All 3 ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 good for agentic (tool-using) tasks?
- In our benchmarks, ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 scores 65.2/100 for agentic workflows. It runs at about 139.6 tok/s, so if you want more speed, swift-1.5-qwen3.8-flash-next-iq3_s is faster (~259.0 tok/s) and still scores well for agentic workflows (88.9/100).
- How fast is ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 for local inference?
- Across 3 community benchmark runs, ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 reaches up to 159.2 tok/s and averages 139.6 tok/s, with the fastest results on Apple M5 Ultra.
- How much memory does ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0 need?
- The leanest observed configuration used about 68.8 GB of memory (quantizations tested: Q2_0).
- Which tools have been used to run ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF:Q2_0?
- Benchmarks were submitted using splash. Results are community-contributed and updated as new runs arrive.