Muse Spark Open

Muse Spark Open is a conversational (chat) large language model attributed to Meta in our catalog. Our database entry for it is intentionally narrow: we record the creator, the model type, and independently measured serving performance, and we do not currently track a published parameter count, context window, or benchmark suite for it. Where a detail is not listed on this page, it means we have not confirmed it — not that the capability is absent. On the performance side, measurements sourced from Artificial Analysis put Muse Spark Open at approximately 104.79 output tokens per second with a time to first token of about 477 milliseconds. Those two figures describe how the model behaves in interactive use: the sub-half-second first-token latency keeps a chat turn from feeling stalled, and the ~105 tokens/second generation rate means a few-hundred-token reply streams out in a handful of seconds. Serving speed depends on the provider, the hardware behind the endpoint, prompt length, and concurrency, so the numbers in the live comparison should be read as a reference point rather than a guarantee. In practice, a model profiled this way is evaluated on whether its latency and throughput fit the interaction pattern you are building — assistant chat, drafting, summarization, and similar text-in/text-out work — rather than on a headline benchmark score. Because our confirmed metadata for Muse Spark Open is limited, readers comparing it against other Meta models or third-party chat models should validate quality on their own prompts and check provider documentation for context limits, modality support, and API features before committing.