Benchmark result
Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL on 2× NVIDIA GeForce RTX 5060 — 118.3 tok/s
Measured with llama.cpp b10884-434ddbbc0 on September 11, 2026.
What the model built
Open full screen →The coding scenario asks for a playable game in a single HTML file. This is exactly what the model returned, unedited; the frame only adds a line that tells this page how tall it is.
How this run compares
1st fastest of 6 runs of this model on 2× NVIDIA GeForce RTX 5060 · Multi-GPU · median 98.1 tok/s.
sort
- 1118.3tok/s79.9llama.cpp · Q2_Kthis run
- 2117.7tok/s77.3llama.cpp · Q2_K
- 398.2tok/s84.0llama.cpp · Q2_K
- 498.0tok/s85.7llama.cpp · Q2_K
- 594.2tok/s68.7llama.cpp · Q2_K
- 693.1tok/s80.4llama.cpp · Q2_K
Reproduce this run
toolllama.cpp b10884-434ddbbc0
modelTiel-Coder-35B-A3B-MTP-UD-Q2_K_XL (Q2_K)
context188,416
thinkingon
samplingtemperature 0.6000000238418579 · top_p 0.8999999761581421 · top_k 0 · min_p 0.05000000074505806 · presence_penalty 0 · repeat_penalty 1
clientv0.4.59+97
Set those in llama.cpp, then:
llm-benchmark benchmark --model "Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL" --tool "llama.cpp"The client prompts for context length and thinking mode, and for the KV cache dtype on Unsloth Studio. Sampling is left at the model default — the values above are what the tool reported using, not overrides the client sent.