Claims and evidence
- Announced Previewed in the Llama 4 announcement as a teacher model that was still training; no release date was given. Primary source[1]page date; Takeaways
- ReleasedRelease date unknown
- Status AnnouncedPreviewed while still in training; Meta said it was not yet releasing the model. No later Meta source reporting a release, weights or API access was found as of 2026-10-01. Primary source[1]Takeaways; opening paragraphs
- Change · Size Nearly two trillion total parameters, with 288B active per token; Meta presents it as taking Llama to new sizes.Compared with Llama 3.1 Primary source[1]Takeaways; Pushing Llama to new sizes: The 2T Behemoth
- Change · Architecture Mixture-of-experts model with 16 experts, trained as the teacher model for distilling the smaller Llama 4 models.Compared with Llama 3.1 Primary source[1]
- Parameters c. 288B active parameters, nearly two trillion total parametersMixture-of-experts model with 16 experts. Meta gives the total only as nearly two trillion (section heading: The 2T Behemoth); the value is that approximation. Primary source[1]Takeaways; Pushing Llama to new sizes: The 2T Behemoth
Lineage
Predecessors
No known predecessor.
Successors
No known successor.
Based on
Not derived from another model.
Variants and derived
- Llama 4 Maverick · 5 April 2025 · derived (distillation)
- Llama 4 Scout · 5 April 2025 · derived (distillation)
Siblings
None recorded.
All ancestors
None.
All descendants
- Llama 4 Maverick · 5 April 2025
- Llama 4 Scout · 5 April 2025
- Muse Spark · 8 April 2026
- Muse Spark 1.1 · 9 July 2026
- Muse Spark 1.2 · 5 August 2026
- Muse Glimmer · 10 August 2026
- Muse Spark 1.3 · 2 September 2026
Variants
No variants recorded in this record.
Related AI Radar coverage
AI Radar coverage starts in June 2026; no coverage linked yet.