04 · Benchmark Analysis
14 / 28
Community models closed the performance gap in eighteen months
Benchmark scores across model categories · MMLU, 5-shot accuracy
75
91.2
Proprietary
Large
84.7
Proprietary
Mid-size
88.4
Open-weight
70B
76.3
Open-weight
13B
82.1
Community
Fine-tuned
69.8
Open-weight
7B