Meta Llama
Meta provides an official API for its Llama family of open-weight models. The Llama series includes some of the most widely used open models in the world, from the efficient Llama 3.3 70B to the massive Llama 3.1 405B and the latest Llama 4 generation with vision support and 1M context.
Getting an API Key
- Visit llama.com/docs/api
- Sign in with your Meta account
- Generate an API key
- Paste the key into AISCouncil under Settings > AI Model > Meta Llama
Meta's official API is pay-as-you-go with no free tier. However, Llama models are also available for free or at low cost through OpenRouter, Together AI, Fireworks AI, Groq, and Cloudflare Workers AI.
API keys are stored locally in your browser (localStorage) and are never included in shared bot URLs.
Supported Models
| Model | Context | Max Output | Input Price | Output Price | Capabilities |
|---|---|---|---|---|---|
| Llama 3.3 70B | 128K | 4K | $0.20/MTok | $0.20/MTok | Tools, code, streaming |
| Llama 3.1 405B | 128K | 4K | $3.00/MTok | $3.00/MTok | Tools, code, streaming |
| Llama 4 Scout | 1M | 4K | $0.15/MTok | $0.50/MTok | Vision, tools, code, streaming |
| Llama 4 Maverick | 1M | 4K | $0.22/MTok | $0.85/MTok | Vision, tools, code, streaming |
Prices are per million tokens (MTok). See llama.com for the latest rates.
- Llama 4 Maverick is the latest flagship -- 1M context, vision support, and strong performance across tasks.
- Llama 4 Scout is the budget-friendly Llama 4 option -- same 1M context window at a lower price.
- Llama 3.3 70B remains a solid all-purpose model at just $0.20/MTok.
- Llama 3.1 405B is the largest open model available -- use it when maximum quality matters.
Available Everywhere
Llama models are open-weight, meaning they are hosted by many inference providers. You can access the same models through:
- OpenRouter -- free tiers available for some Llama models
- Together AI -- Turbo-optimized variants
- Fireworks AI -- low-latency serving with 16K max output
- Groq -- ultra-fast LPU inference, free tier
- Cloudflare Workers AI -- edge inference, free 10K neurons/day
The Meta official API offers the most direct access and competitive pricing, but third-party providers may offer benefits like free tiers, lower latency, or higher max output tokens.
Configuration
Select Meta Llama as the provider when creating a bot profile. The app connects to api.llama.com/v1 using the OpenAI-compatible format with Bearer authentication.
Tips for Best Results
- Compare Meta's pricing with third-party providers. The same models may be cheaper (or free) on OpenRouter, Groq, or Cloudflare.
- Use Llama 4 Scout/Maverick for multimodal tasks. Both support vision input, allowing you to analyze images alongside text.
- Try Llama 3.1 405B for maximum open-model quality. It is the largest publicly available model and competes with closed models on many benchmarks.