Meta

Maker of the open-weight Llama family.

Meta releases the Llama family of open-weight models, available across a wide range of inference providers including AWS Bedrock, Groq, and Fireworks.

ai.meta.com

28 model routes14 providersEU + US regions2 Sep 2026 latest release

Models from Meta

Model	Context	Zero data retention	Training	Region	Provider
Muse Spark 1.3 Meta	1M	Not by default	No	Multi	Meta
Muse Spark 1.3 Contributor Meta	1M	Not by default	Yes	Multi	Meta
Muse Spark 1.2 Meta	1M	Not by default	No	Multi	Meta
Muse Glimmer Meta	128K	Zero data retention	No	EU	Melious
Llama 3.3 70B Meta	131K	Not by default	No	EU	Nebius
Llama 3.3 70B Meta	131K	Zero data retention	No	US	Novita
Llama 3.3 70B Instruct Meta	131K	Zero data retention	No	US	Infercom
Llama 3.3 70B Meta	128K	Zero data retention	No	EU	Berget
Llama 3.3 70B Meta	128K	Zero data retention	No	EU	Regolo
Llama 3.3 70B Instruct Meta	128K	Zero data retention	No	EU	evroc
Llama 3.1 8B Instruct Meta	131K	Zero data retention	No	US	DeepInfra
Llama 3.1 8B Instruct Meta	16K	Zero data retention	No	US	Novita
Llama 3.1 70B Instruct Meta	131K	Zero data retention	No	US	DeepInfra
Llama 3.2 11B Vision Instruct Meta	131K	Zero data retention	No	US	DeepInfra
Llama 4 Maverick 17B 128E Instruct FP8 Meta	1M	Zero data retention	No	US	DeepInfra
Llama 4 Maverick 17B 128E Instruct FP8 Meta	1M	Zero data retention	No	US	Novita
Llama 4 Scout 17B 16E Instruct Meta	328K	Zero data retention	No	US	DeepInfra
Llama 4 Scout 17B 16E Instruct Meta	131K	Zero data retention	No	US	Novita
Llama 4 Maverick 17B 128E Instruct FP8 Meta	128K	Not by default	No	US	DigitalOcean
Llama 3.3 70B Instruct Turbo Meta	131K	Zero data retention	No	US	DeepInfra
Llama 3.3 70B Meta	128K	Not by default	No	US	Groq
Llama 3.1 70B Instruct Turbo Meta	131K	Zero data retention	No	US	DeepInfra
Llama 3.1 8B Instruct Turbo Meta	131K	Zero data retention	No	US	DeepInfra
Llama 3.1 Nemotron Ultra 253B v1 Meta	131K	Not by default	No	EU	Nebius
Llama 3.1 8B Instant Meta	128K	Not by default	No	US	Groq
Llama Guard 4 12B Meta	164K	Zero data retention	No	US	DeepInfra
Muse Glimmer 30B Meta	131K	Zero data retention	No	Multi	EmpirioLabs
Muse Glimmer 30B Meta	131K	Zero data retention	No	Multi	Fireworks

Release timeline

Every dated Meta release, newest first — 7 of 21 run on the Opper gateway today. Releases without a published date aren't listed. See the Meta release timeline.

5 releases in the last 12 months25 days median gap5–92 days range

Not currently available

Meta models without a live gateway route right now. Their pages stay up for reference — contact us if you'd like one added or brought back.

Llama 3 8B InstructNot currently offered Llama 3.1 8BNot currently offered Llama 3.2 1B InstructNot currently offered Llama 3.2 3B InstructNot currently offered Llama3 70B InstructNot currently offered