Clairic is Cloudflare’s extraction model that answers in a single pass without generating output tokens. Send text and up to four images with typed fields, and get back extracted values, computed values, estimates with 80% intervals, and forecasts, each with a confidence score. It’s in preview on Workers AI and the REST API. Cloudflare says open weights are planned. It charges only for input tokens.
Nemotron 3.5 Bolt 24GB is NVIDIA’s model for running everyday agentic tasks locally on NVIDIA RTX and DGX Spark hardware. It compresses Nemotron 3.5 Super VL from 120B total, 12B active to 74B total, 9B active, sized to fit a 24GB memory budget. NVIDIA reports 84 tokens per second on RTX Spark. Text and vision. Checkpoints, GGUF files, and a speculative-decoding drafter are on Hugging Face.
Nemotron 3.5 Super VL is NVIDIA’s open, customizable multimodal model for specialized agents. It is a 120B mixture-of-experts model with 12B active parameters that handles text, image, document, and video input, plus reasoning and tool use, in one model. NVIDIA claims up to 5x higher agentic throughput on a single GPU than open models its size. Checkpoints, training recipes, and datasets are on Hugging Face, with a NIM container.
