Three new open-weight models from US companies give builders more choice

Reflection Beam, Meta’s Muse Spark Open, and NVIDIA Nemotron 3.5 Super VL have launched on Nebius Token Factory, expanding the selection of open models for coding, reasoning, tool use, and visual workflows through a shared OpenAI-compatible API. Beam is an Apache 2.0 text-only mixture-of-experts model focused on coding and reasoning, with adjustable reasoning effort, 501 billion total parameters, and relatively low token pricing. Muse Spark Open is Meta’s open-weights multimodal coding model, with 816 billion total parameters, designed for multi-file debugging, refactoring, codebase understanding, and tool use with text, image, and video inputs. Nemotron 3.5 Super VL is a smaller visual-language model with 120 billion total parameters and 12 billion active parameters, intended for document understanding, web research, code validation, and customer-service tasks requiring visual evidence. The release encourages developers to evaluate models against representative production tasks using measures such as output quality, tool selection, latency, retries, and cost per successful run, rather than relying solely on benchmark rankings.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.