Skip to main content

Meta Llama API

Access Meta's Llama models directly through the official Llama API.

Configuration

FieldRequiredExample
Application NameYesmy-llama-app
Model NameYesLlama-4-Maverick-17B-128E-Instruct
API KeyYesYour Llama API key

Getting Your API Key

  1. Go to Meta Llama Developer
  2. Sign up for API access
  3. Navigate to your dashboard
  4. Copy your API key
ModelDescription
Llama-4-Maverick-17B-128E-InstructFlagship MoE (400B total, 17B active), 1M context, 128 experts
Llama-4-Scout-17B-16E-InstructBest multimodal in its class, 10M context, fits single H100
Llama-4-Behemoth288B active params, 2T total (teacher model, still training)
Llama-3.3-70B-InstructHigh-quality text-only model

Note: Llama 4 models are natively multimodal with mixture-of-experts architecture, supporting 12 languages for image/text understanding.

For the complete list of available models, visit the Llama API Models Documentation.

Advanced Parameters

ParameterTypeDescription
max_completion_tokensintegerMaximum tokens
temperaturefloatRandomness
top_pfloatNucleus sampling
top_kintegerTop-k sampling
repetition_penaltyfloatPenalize repetition

Setup Steps

  1. Navigate to AI Applications → New Application
  2. Select Model Providers tab
  3. Click Meta Llama API
  4. Enter your Application Name and API Key
  5. Select your desired Model Name
  6. Configure Advanced Settings (optional)
  7. Click Test Response to verify (optional)
  8. Review and submit