LLAMA API
Llama 4 and 3, hosted and ready.
Call Meta's open-weight Llama 4 Scout, Llama 3.3 and Llama 3.1 through an OpenAI-compatible endpoint, with no infrastructure to run.
Hand this to your agent
Add a Runcrate model to this project: run curl -fsSL"https://runcrate.ai/agents.md?intent=build"and follow it.
Your agent asks for an API key. Create a key
Quick start
Hand it to your agent. Or read the request it sends.
Your agent writes and runs it. The same key works for every model and every GPU.
Available models
Models you can call today. Billed prices, as charged.
| Model | Maker | Price | Detail |
|---|---|---|---|
| meta-llama/Llama-4-Scout-17B-16E-Instruct | Meta | $0.105 in, $0.315 out per 1M tokens | 17B MoE, 128K context, newest |
| meta-llama/Llama-3.3-70B-Instruct-Turbo | Meta | $0.105 in, $0.336 out per 1M tokens | 70B dense, strong reasoning |
Why Runcrate
Built for production. One key, one balance.
Llama 4 MoE
Llama 4 Scout uses 17B active parameters from a 16-expert mixture, delivering strong quality with efficient inference.
Open weights
All Llama models are open-weight. Use the API for convenience, or download weights to self-host on Runcrate GPU instances for full control.
Full model range
From the lightweight 8B model for fast prototyping to the 70B powerhouse for maximum quality. Pick the right size for your latency and cost budget.
128K context
Llama 4 Scout supports 128K tokens of context, enough for entire codebases, lengthy documents, or extended multi-turn conversations.
FAQ
Common questions.
Related