Model benchmarks
ds-v4-mtp-q4km local LLM performance
As of November 2026, ds-v4-mtp-q4km runs at up to 8.9 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
Unknown
Peak speed
8.9 tok/s
Average speed
8.9 tok/s
Avg PP
90.9 tok/s
Min memory
7.3 GB
Max context
8,192 tokens
Avg output / run
13,108 tokens
Avg runtime / run
24m 39s
Avg quality
71.0
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 | 68.2 | 71.0 | 73.9 |
| Agent Workflow | 70.9 | 77.1 | 83.3 |
| Code Generation | 61.4 | 64.6 | 67.8 |
| Role Play & Narrative | 62.5 | 64.4 | 66.3 |
| Research & Analysis | 74.3 | 78.0 | 81.8 |
Performance by hardware and tool
Every hardware/tool/quantization combination ds-v4-mtp-q4km 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 680M | llama.cpp | — | 8.9 tok/s | 8.9 tok/s | 7.3 GB | 8,192 tokens | 71.0 | 2 |
Benchmark runs
All 2 ds-v4-mtp-q4km runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is ds-v4-mtp-q4km good for coding?
- In our benchmarks, ds-v4-mtp-q4km scores 64.6/100 for coding. It runs at about 8.9 tok/s, so if you want more speed, Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp is faster (~43.5 tok/s) and still scores well for coding (86.6/100).
- Is ds-v4-mtp-q4km good for agentic (tool-using) tasks?
- In our benchmarks, ds-v4-mtp-q4km scores 77.1/100 for agentic workflows. It runs at about 8.9 tok/s, so if you want more speed, Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW is faster (~18.9 tok/s) and still scores well for agentic workflows (92.2/100).
- How fast is ds-v4-mtp-q4km for local inference?
- Across 2 community benchmark runs, ds-v4-mtp-q4km reaches up to 8.9 tok/s and averages 8.9 tok/s, with the fastest results on AMD Radeon 680M.
- How much memory does ds-v4-mtp-q4km need?
- The leanest observed configuration used about 7.3 GB of memory.
- Which tools have been used to run ds-v4-mtp-q4km?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.