Provider

Meta

Meta is best evaluated through its open-weight Llama family and its newer Muse models, served by many providers at very different prices.

summary

Meta Platforms, headquartered in Menlo Park, California, develops the Llama family of open-weight language models and, more recently, the Muse models. Llama models range from small 1B and 3B versions to Llama 4 Maverick and Scout with very long context, and they are served by many inference providers, which makes price and speed comparisons especially useful. The Muse models (Muse Spark, Muse Wonder, Muse Glimmer) add reasoning, tool calling, web search and image input with context windows up to about one million tokens. Meta also publishes safety classifiers such as Llama Guard and Prompt Guard. On Eden AI, all of these are available through one API key, with each model's providers, prices and regions listed side by side.

What is Meta?

Meta Platforms is an American technology company headquartered in Menlo Park, California. Its AI research and product teams develop the Llama family of language models, released with open weights so that companies can run them with the provider of their choice or on their own infrastructure. Meta has since added the Muse family, aimed at reasoning and agentic use cases.

Meta models available on Eden AI

Muse Spark (1.1, 1.2, 1.3) and Muse Wonder 1.0: reasoning, function calling, web search and image input, with context windows of about one million tokens.
Muse Spark Open and Muse Glimmer 30B: open versions of the Muse family with reasoning, function calling and image input.
Llama 4 Maverick and Llama 4 Scout: multimodal models (text and image input) with function calling and long context.
Llama 3.3 70B and Llama 3.1 (8B, 70B): widely deployed text models with function calling, served by many providers.
Llama 3.2 (1B, 3B, 11B Vision): small and vision models for lightweight or edge use cases.
Llama Guard 3 and 4, Prompt Guard 2: safety classifiers that check prompts and responses.
H2: When to choose Meta models
You want open-weight models you can also run elsewhere, without lock-in.
You need a model available from many providers, to compare prices and keep a fallback.
You want reasoning and tool calling with very long context (Muse family).
You need a safety layer in front of another model (Llama Guard, Prompt Guard).

How to use Meta models with Eden AI

Call any Meta model with the Eden AI chat completions API and your Eden AI API key. Send the model name alone to let Eden AI choose the provider, or pin a provider with the endpoint ID shown on the model page. Each model page lists every provider serving it, with prices, regions and measured speed.

‍