Model benchmarks
Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL local LLM performance
As of October 2026, Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL runs at up to 14.1 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
35B
Peak speed
14.1 tok/s
Average speed
14.0 tok/s
Avg PP
121.2 tok/s
Min memory
13.3 GB
Max context
8,192 tokens
Avg output / run
4,702 tokens
Avg runtime / run
5m 34s
Avg quality
75.4
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 | 70.0 | 75.4 | 80.8 |
| Agent Workflow | 62.2 | 72.1 | 82.0 |
| Code Generation | 72.6 | 76.3 | 80.0 |
| Role Play & Narrative | 76.3 | 80.1 | 83.9 |
| Research & Analysis | 69.0 | 73.1 | 77.3 |
Performance by hardware and tool
Every hardware/tool/quantization combination Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL 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 |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon 680M | llama.cpp | Q2_K | 14.1 tok/s | 14.0 tok/s | 13.3 GB | 8,192 tokens | 75.4 | 2 |
Benchmark runs
All 2 Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL good for coding?
- In our benchmarks, Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL scores 76.3/100 for coding. It runs at about 14.0 tok/s, so if you want more speed, tensorfold-qwen38-27b is faster (~25.0 tok/s) and still scores well for coding (84.3/100).
- Is Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL good for agentic (tool-using) tasks?
- In our benchmarks, Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL scores 72.1/100 for agentic workflows. It runs at about 14.0 tok/s, so if you want more speed, muse-glimmer:latest is faster (~33.3 tok/s) and still scores well for agentic workflows (91.6/100).
- How fast is Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL for local inference?
- Across 2 community benchmark runs, Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL reaches up to 14.1 tok/s and averages 14.0 tok/s, with the fastest results on AMD Radeon 680M.
- How much memory does Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL need?
- The leanest observed configuration used about 13.3 GB of memory (quantizations tested: Q2_K).
- Which tools have been used to run Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.