Phi-5
Phi-5 is Microsoft’s small language model family, demonstrating that compact 7B-14B models can match much larger systems on many benchmarks.
News & Notes
35 posts
Nemotron 4
Nemotron 4 is NVIDIA’s open-weights model family optimized for synthetic data generation and reasoning. Available in multiple sizes.
Llama 4 Maverick
Llama 4 Maverick is Meta’s flagship open-weights model with strong reasoning and multilingual capabilities. Apache-compatible license for self-hosting.
Llama 4 Scout
Llama 4 Scout is a smaller efficient variant of the Llama 4 family, designed for single-GPU inference with multimodal capabilities.
DeepSeek V4 Pro
DeepSeek V4 Pro is a flagship open-weights model with 1.6T total parameters (MoE) designed for long-context reasoning, coding, and agentic…