Machine
Apple M1 Max · Apple M1 Max #55672f1c
63 runs of 18 models since Nov 2, 2026. Last run Nov 4, 2026.
63
runs
18
models
55
configs
8
configs run 2+ times
Hardware
- GPU
- Apple M1 Max 64 GB
- CPU
- Apple M1 Max 10 cores
- RAM
- 64 GB
- OS
- macOS 27.2
- Tools
- LM Studio 63
- Client
- 0.4.98
Models on this machine
One dot per model, at its best config. Up is better answers, right is faster. Time to finish counts every token the model wrote, thinking included, so a model that talks a lot ends up further left.
Best trade-off: quality 60+, and nothing else here is both faster and betterOther models · dot size = number of runs
Medians of each model's best config, 18 models.
Best config per model
sort
Runs with the same model, quant, tool, context, thinking and KV cache count as one config. Numbers are the median of its runs, and each dot is one run on the 0–100 quality scale. ★ marks the best config with at least 2 rated runs.
configrunsqualityeach runspeed
qwen/qwen3.6-35b-a3b
11 runs · 8 configs
★ best6bitLM Studio8k ctxthinking default
×2 runs
82.079.9–84.1
steady
57.7 tok/s
47.5–67.9
google/gemma-4-12b-qat
1 run · 1 config
google/gemma-4-26b-a4b
1 run · 1 config
qwen_qwen3.5-4b@q8_0
2 runs · 2 configs
qwen3.5-4b@q4_k_m
2 runs · 2 configs
qwen3.5-4b@q6_k
2 runs · 2 configs
qwen3.5-4b@q8_0
1 run · 1 config
google/gemma-4-e2b
12 runs · 12 configs
qwen/qwen3.5-9b
2 runs · 2 configs
google/gemma-4-e4b
6 runs · 4 configs
★ best4bitLM Studio64k ctxthinking default
×2 runs
65.962.9–69.0
steady
46.6 tok/s
45.3–48.0
Compare with another machine
Run history
Nov 4, 2026
21:43gemma-4-e2bLM Studio71.3 tok/s61.821:37gemma-4-12bLM Studio29.6 tok/s68.921:21gemma-4-12bLM Studio22.7 tok/s65.320:23gemma-4-e4bLM Studio48.3 tok/s66.720:19gemma-4-e4bLM Studio45.8 tok/s60.420:13gemma-4-12b-qatLM Studio27.8 tok/s74.716:12qwen3.6-35b-a3bLM Studio68.7 tok/s82.115:02qwen3.6-35b-a3bLM Studio53.9 tok/s81.014:43qwen3.8-27bLM Studio16.0 tok/s83.314:25qwen3.8-27bLM Studio15.4 tok/s87.412:20qwen3.6-35b-a3bLM Studio59.8 tok/s79.512:13qwen3.6-35b-a3bLM Studio69.4 tok/s76.001:31gemma-4-e2bLM Studio76.4 tok/s68.201:27gemma-4-e2bLM Studio78.6 tok/s68.101:25gemma-4-e2bLM Studio78.4 tok/s39.501:23gemma-4-e2bLM Studio78.8 tok/s57.301:19gemma-4-e2bLM Studio78.9 tok/s64.701:14gemma-4-e2bLM Studio78.5 tok/s64.401:11gemma-4-e2bLM Studio78.7 tok/s61.701:04gemma-4-e2bLM Studio56.8 tok/s69.401:00gemma-4-e2bLM Studio74.4 tok/s63.2Nov 3, 2026
22:10qwen3.5-4b@q8_0LM Studio45.4 tok/s69.921:53qwen_qwen3.5-4b@q8_0LM Studio67.9 tok/s66.721:46qwen_qwen3.5-4b@q8_0LM Studio69.0 tok/s73.021:26qwen_qwen3.5-4bLM Studio52.6 tok/s63.420:37qwen3.5-4b@q6_kLM Studio45.0 tok/s72.320:28qwen3.5-4b@q4_k_mLM Studio49.1 tok/s72.920:21qwen3.5-4b@q6_kLM Studio46.7 tok/s58.320:13qwen3.5-4b@q4_k_mLM Studio52.2 tok/s62.019:08qwen3.8-27bLM Studio17.0 tok/s86.418:45qwen3.8-27bLM Studio16.9 tok/s82.318:07qwen3.8-27bLM Studio17.4 tok/s77.917:45qwen3.8-27bLM Studio17.8 tok/s57.017:31qwen3.8-27bLM Studio16.7 tok/s73.915:22gemma-4-e4bLM Studio47.0 tok/s60.915:18gemma-4-e2bLM Studio77.6 tok/s62.315:12gemma-4-e2bLM Studio78.9 tok/s65.315:09gemma-4-e4bLM Studio48.0 tok/s62.915:03gemma-4-12bLM Studio29.1 tok/s64.514:15qwen3.5-9bLM Studio48.5 tok/s67.114:03qwen3.5-4b-mlxLM Studio78.5 tok/s60.713:53qwen3.5-4b-mlxLM Studio84.2 tok/s54.412:06qwen3.8-9b-distill-mlx@4bitLM Studio49.9 tok/s63.111:33qwen3.8-27bLM Studio12.5 tok/s66.310:57qwen3.6-35b-a3bLM Studio86.0 tok/s80.610:46qwen3.6-35b-a3bLM Studio66.3 tok/s78.710:18qwen3.6-35b-a3b×2LM Studio57.7 tok/s82.009:47qwen3.6-35b-a3bLM Studio48.2 tok/s81.201:24qwen3.8-27bLM Studio14.8 tok/s67.101:03gemma-4-12bLM Studio26.1 tok/s62.900:38gemma-4-e4bLM Studio45.3 tok/s69.000:30qwen3.8-9b-distill-mlx@4bitLM Studio42.2 tok/s64.100:18qwen3.6-35b-a3bLM Studio58.2 tok/s81.3Nov 2, 2026
23:56gemma-4-e4bLM Studio45.5 tok/s60.623:43gemma-4-26b-a4bLM Studio48.6 tok/s73.723:36qwen3.6-35b-a3bLM Studio59.7 tok/s75.923:30qwen3.8-9b-distill-mlx@8bitLM Studio32.9 tok/s65.723:17qwen3.8-9b-distill-mlx@4bitLM Studio41.1 tok/s67.723:08qwen3.8-2b-distill-mlx@4bitLM Studio123 tok/s23.523:02qwen3.8-27bLM Studio13.9 tok/s71.622:37qwen3.5-9bLM Studio43.0 tok/s56.622:19qwen3.8-4b-distill-mlxLM Studio66.7 tok/s61.9