Model benchmarks
qwen3.8-27b-splash local LLM performance
As of September 2026, qwen3.8-27b-splash runs at up to 72.7 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
72.7 tok/s
Average speed
71.9 tok/s
Avg PP
803.0 tok/s
Min memory
n/a
Max context
131,072 tokens
Avg output / run
16,755 tokens
Avg runtime / run
4m 31s
Avg quality
83.4
Benchmark runs
2
GPUs tested
1
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 | 81.8 | 83.4 | 85.0 |
| Agent Workflow | 73.6 | 77.6 | 81.6 |
| Code Generation | 77.3 | 80.0 | 82.7 |
| Role Play & Narrative | 90.7 | 92.0 | 93.4 |
| Research & Analysis | 83.0 | 83.9 | 84.9 |
Performance by hardware and tool
Every hardware/tool/quantization combination qwen3.8-27b-splash has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| Apple M5 Max | LM Studio | — | 72.7 tok/s | 71.9 tok/s | n/a | 131,072 tokens | 83.4 | 2 |
Benchmark runs
All 2 qwen3.8-27b-splash runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is qwen3.8-27b-splash good for coding?
- In our benchmarks, qwen3.8-27b-splash scores 80.0/100 for coding. It runs at about 71.9 tok/s, so if you want more speed, ornith-ai/Ornith-1.5-35B-A3B-GGUF is faster (~96.3 tok/s) and still scores well for coding (83.4/100).
- Is qwen3.8-27b-splash good for agentic (tool-using) tasks?
- In our benchmarks, qwen3.8-27b-splash scores 77.6/100 for agentic workflows. It runs at about 71.9 tok/s, so if you want more speed, Ornith-1.5-35B-A3B-BigBang-oQ8e-mtp is faster (~102.7 tok/s) and still scores well for agentic workflows (91.1/100).
- How fast is qwen3.8-27b-splash for local inference?
- Across 2 community benchmark runs, qwen3.8-27b-splash reaches up to 72.7 tok/s and averages 71.9 tok/s, with the fastest results on Apple M5 Max.
- Which tools have been used to run qwen3.8-27b-splash?
- Benchmarks were submitted using LM Studio. Results are community-contributed and updated as new runs arrive.