Phi-3.5-MoE Instruct vs Seed 1.5-VL

Phi-3.5-MoE Instruct

Microsoft

Phi-3.5-MoE-instruct is a sparse mixture-of-experts model from Microsoft's Phi research team, released in August 2024 with 42 billion total parameters across 16 experts and approximately 6.6 billion active parameters per forward pass. The model applies Microsoft's small-data, high-quality training philosophy — developed across earlier Phi generations — to a MoE architecture, targeting reasoning quality comparable to much larger dense models at a fraction of the inference compute. Released under the MIT license, it was notable in the research community for demonstrating that MoE efficiency gains could be realized at smaller total parameter counts than typical large-scale MoE deployments.

ByteDance

Seed1.5-VL, released by ByteDance Seed on May 15, 2025, is a vision-language foundation model composed of a 532M-parameter vision encoder and a Mixture-of-Experts language model with 20 billion active parameters. It was pretrained on over 3 trillion multimodal tokens and achieved state-of-the-art performance on 38 out of 60 public VLM benchmarks at release. Seed1.5-VL targets complex visual reasoning, OCR, video comprehension, 3D spatial understanding, and multimodal agentic tasks.