Model benchmarks

/home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf local LLM performance

As of August 2026, /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf runs at up to 36.1 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

llama.cppQ6_K
ShareRedditX

Model size

35B

Peak speed

36.1 tok/s

Average speed

35.6 tok/s

Min memory

17.1 GB

Max context

85,248 tokens

Avg output / run

28,982 tokens

Avg runtime / run

13m 13s

Avg quality

86.7

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.

TaskP5 (low)AvgP95 (high)
Overall85.986.787.5
Agent Workflow90.691.191.5
Code Generation83.484.986.5
Role Play & Narrative80.085.390.5
Research & Analysis84.785.586.3

Performance by hardware and tool

Every hardware/tool/quantization combination /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf has been benchmarked on, ranked by peak token generation speed. Last updated August 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
AMD Radeon RX 5600 OEM/5600 XT / 5700/5700 XTllama.cppQ6_K36.1 tok/s35.6 tok/s17.1 GB85,248 tokens86.72

Frequently asked questions

Is /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf good for coding?
In our benchmarks, /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf scores 84.9/100 for coding, currently the best-scoring model for coding that runs on consumer hardware (≤24 GB VRAM). It runs at about 35.6 tok/s, so if you want more speed, Ornith-1.5-35B-A3B-oQ4e-mtp is faster (~111.8 tok/s) and still scores well for coding (81.8/100).
Is /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf good for agentic (tool-using) tasks?
In our benchmarks, /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf scores 91.1/100 for agentic workflows, among the top 2 for agentic workflows on consumer hardware (≤24 GB VRAM). It runs at about 35.6 tok/s, so if you want more speed, openai/gpt-oss-20b:2 is faster (~154.8 tok/s) and still scores well for agentic workflows (87.2/100).
How fast is /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf for local inference?
Across 2 community benchmark runs, /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf reaches up to 36.1 tok/s and averages 35.6 tok/s, with the fastest results on AMD Radeon RX 5600 OEM/5600 XT / 5700/5700 XT.
How much memory does /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf need?
The leanest observed configuration used about 17.1 GB of memory (quantizations tested: Q6_K).
Which tools have been used to run /home/polaris/models/Tiel-Coder-35B-A3B-MTP-UD-Q6_K_XL.gguf?
Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.