Fireworks AI tokens & top-ups

About the service

Fireworks AI is a fast-inference platform for open models such as Llama, Qwen, DeepSeek and Mixtral, tuned for low latency and high throughput. It supports function calling, JSON-schema structured output, multimodal and audio models, image generation, and fine-tuning with deployment of your own LoRA. Dedicated deployments are available when you need predictable performance.