L

Meta Llama API Pricing

Updated April 2026 · 5 models

Meta's Llama models are open-weight, meaning you can self-host for free or pay for managed API access. Llama 4 Maverick and Scout represent the frontier of open-source model performance. Prices below reflect hosted API pricing (via together.ai and similar providers).

🔓Open-weight models
🚀Llama 4: frontier performance
💰Self-host for $0 API cost
Llama 3.1 8B: ultra-fast & cheap
Llama 4 Maverick
Meta's powerful Llama 4 model balancing performance and cost with 1M context
$0.5
input /1M
$1.1
output /1M
Llama 4 Scout
Meta's latest efficient model with a massive 10M token context window at extremely low cost
$0.17
input /1M
$0.17
output /1M
Llama 3.1 405B
Meta's largest open-source model — frontier-class intelligence available via third-party API providers
$3.5
input /1M
$3.5
output /1M
Llama 3.3 70B
Meta's large open-source model with strong reasoning. Great value via providers like Together AI or Fireworks.
$0.23
input /1M
$0.4
output /1M
Llama 3.1 8B
Meta's efficient open-source model. Cheapest option for high-volume tasks via third-party API providers.
$0.02
input /1M
$0.05
output /1M

Compare Meta Llama vs Other Providers

vs OpenAI GPTvs Anthropic Claudevs DeepSeekLlama 4 Maverick vs Claude SonnetCheapest LLM APIs →