Meta Muse Wonder: Text and Image Input Model
Meta released Muse Wonder on October 12, 2026. This model accepts text and image inputs to generate text outputs. Open weights are not confirmed.
Release Details and Modalities
Meta released Muse Wonder on October 12, 2026. The announcement identifies the model as accepting both text and image inputs. The output modality is specified as text only. The source event date is distinct from this article’s publication date; the historical record establishes the release occurred on the stated day in 2026.
- Provider: Meta
- Input Modalities: Text, Image
- Output Modities: Text
- Open Weights: Not confirmed
Scope of Verified Information
The available documentation confirms the existence of the model and its basic input-output structure. It does not provide details on context window size, maximum output tokens, or pricing structures. The field for open weights is marked as false, which records that open weights are not confirmed. This status does not establish that local installation is impossible, nor does it mandate API access or define any specific hosting arrangement. No architectural details, such as parameter counts or computational depth mechanisms, are present in the source facts.
Evaluation Procedure for Engineers
To assess the model's capabilities within the verified scope, engineers should design a controlled test suite. First, prepare a dataset containing paired text and image inputs with known expected textual responses. Second, measure the model's output accuracy by comparing generated text against these ground truth values. Third, validate the system's ability to process mixed-modality inputs without error. Finally, compare the results against internal baselines to determine if the model meets specific project requirements. These steps focus on observable behavior rather than inferring internal architecture or unverified benefits.
Primary sources
Frequently asked questions
Does the source confirm that Muse Wonder has open weights?
No. The source indicates that open weights are not confirmed. This fact does not imply that local installation is prohibited or that API usage is the only option; it simply means the availability of open weights is not established in the provided data.
What specific performance metrics are provided for Muse Wonder?
The source does not provide specific performance metrics, benchmark scores, or latency figures. It only confirms the input modalities (text and image) and the output modality (text). Any evaluation of performance must be conducted through independent testing by the user.