Qwen 3.5 122B

Meta researchers developed Proteus, a system that uses AI agents to generate specialized GPU kernels, achieving 1.8–5.2× faster inference for Qwen 3.5 122B models compared to the best kernels in vLLM.…

Pokee AI's Isaac 28B v0 model is the only one in a six-model comparison to return usable scores at every context length up to 10M tokens, achieving 93.3 on the RULER benchmark at 10M, while competitor…