Benchmark result
mellum2.1-12b-a2.5b-thinking on Apple M5 Max — 98.3 tok/s
Measured with LM Studio on October 9, 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 3 runs of this model on Apple M5 Max · median 96.8 tok/s.
sort
- 198.3tok/s64.1LM Studio · BF16this run
- 296.8tok/s70.9LM Studio · BF16
- 396.2tok/s68.8LM Studio · BF16
Reproduce this run
toolLM Studio
modelmellum2.1-12b-a2.5b-thinking (BF16)
context65,536
thinkingon
clientv0.4.85+101
Set those in LM Studio, then:
llm-benchmark benchmark --model "mellum2.1-12b-a2.5b-thinking" --tool "LM Studio"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.