Benchmark result
Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q5_K_XL on AMD Radeon 8050S — 51.5 tok/s
Measured with llama.cpp b11223-4da633776 on September 30, 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
5th fastest of 5 runs of this model on AMD Radeon 8050S · median 54.1 tok/s.
sort
- 154.8tok/s85.3llama.cpp · Q5_K
- 254.8tok/s85.2llama.cpp · Q5_K
- 354.1tok/s85.1llama.cpp · Q5_K
- 452.8tok/s83.3llama.cpp · Q5_K
- 551.5tok/s86.6llama.cpp · Q5_Kthis run
Reproduce this run
toolllama.cpp b11223-4da633776
modelCyber-Tiel-Coder-35B-A3B-MTP-UD-Q5_K_XL (Q5_K)
context65,536
samplingtemperature 0.6000000238418579 · top_p 0.949999988079071 · top_k 20 · min_p 0 · presence_penalty 0 · repeat_penalty 1
clientv0.4.85+101
Set those in llama.cpp, then:
llm-benchmark benchmark --model "Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q5_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.