Release tracker / Meta
Meta model releases
Every Meta model launch we have recorded, newest first: Muse Spark, Llama. Each one links to its source, and to how it ranks on our benchmarks where we have tested it.
- Latest
- Muse Spark 1.3, 2 Sep 2026
- Launches
- 11
- Since
- 18 Apr 2024
- Since last launch
- 40 days
Meta's latest model
-
#6 of 22 overall$1.25 in · $4.25 out per MReference source
Muse Spark 1.3 came out 40 days ago. Spotted a launch we're missing? Tell us.
How often Meta ships
Lines with at least three launches in the last two years. Amber: longer since the last launch than the line's average gap. Grey: no launch for 90 days or more, marked quiet.
| Line | Size | Launches | Average gap | Latest | Last launch |
|---|---|---|---|---|---|
| Muse Spark | Large | 4 | 49 days | Muse Spark 1.32 Sep 2026 | 40 days ago |
Every Meta release
-
#6 of 22 overall$1.25 in · $4.25 out per MReference source
-
$1.25 in · $4.25 out per MReference source
-
$1.25 in · $4.25 out per MReference source
-
$0.18 in · $0.59 out per MEpoch AI source
-
$0.27 in · $0.85 out per MEpoch AI source
-
$0.27 in · $0.85 out per MProvider source
-
$0.59 in · $0.79 out per MReference source
-
$0.05 in · $0.33 out per MProvider source
-
Llama 3
Questions
- What is Meta's latest model?
- Muse Spark 1.3, released 2 Sep 2026 (40 days ago). It is the newest of the 11 Meta launches we have recorded since 18 Apr 2024.
- When will Meta release its next model?
- Meta has not announced its next model. It usually releases a new Muse Spark model about every 49 days; the last was Muse Spark 1.3, 40 days ago.
- How often does Meta release new models?
- Over the last two years: a new Muse Spark model about every 49 days (4 launches).
Other providers: OpenAI, Mistral AI, Alibaba, Anthropic, Google, DeepSeek, xAI, Z.ai, Moonshot AI, Microsoft, MiniMax, Xiaomi, Amazon, NVIDIA, StepFun. Or see every launch on one timeline.
A launch is the first day a model was publicly available, previews included; a family's sizes released together count once, and later dated updates of a model are left out. Main lines are checked by hand against the provider's own changelog or announcement where we could read it ("Provider"), otherwise a reference page ("Reference"); the rest come from Epoch AI's model data (CC BY 4.0).