Model benchmarks
qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp local LLM performance
As of November 2026, qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp runs at up to 48.9 tok/s for local inference (best of 4 community benchmark runs across 0 GPUs).
Model size
27B
Peak speed
48.9 tok/s
Average speed
46.8 tok/s
Avg PP
409.0 tok/s
Min memory
n/a
Max context
65,536 tokens
Avg output / run
14,232 tokens
Avg runtime / run
4m 58s
Avg quality
80.4
Benchmark runs
4
GPUs tested
0
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 4 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 | 77.6 | 80.4 | 82.3 |
| Agent Workflow | 72.1 | 79.6 | 86.8 |
| Code Generation | 74.8 | 76.9 | 79.9 |
| Role Play & Narrative | 76.8 | 82.5 | 89.5 |
| Research & Analysis | 79.0 | 82.5 | 85.5 |
Performance by hardware and tool
Every hardware/tool/quantization combination qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp 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 |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon Graphics | llama.cpp | — | 48.9 tok/s | 46.8 tok/s | n/a | 65,536 tokens | 80.4 | 4 |
Benchmark runs
All 4 qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp good for coding?
- In our benchmarks, qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp scores 76.9/100 for coding. It runs at about 46.8 tok/s, so if you want more speed, turbo is faster (~124.1 tok/s) and still scores well for coding (85.3/100).
- Is qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp good for agentic (tool-using) tasks?
- In our benchmarks, qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp scores 79.6/100 for agentic workflows. It runs at about 46.8 tok/s, so if you want more speed, granite-4.2-8b-Q4_K_L is faster (~71.6 tok/s) and still scores well for agentic workflows (92.9/100).
- How fast is qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp for local inference?
- Across 4 community benchmark runs, qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp reaches up to 48.9 tok/s and averages 46.8 tok/s, with the fastest results on AMD Radeon Graphics.
- Which tools have been used to run qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_S-mtp?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.