Serverless Modes

Fireworks Serverless offers four modes:

Standard is the default mode. No service_tier parameter is needed.
Priority tier is for workloads that require higher reliability during peak traffic.
Fast is for workloads that require higher speeds.
Ultra-fast (Preview) for workloads requiring extremely fast end-to-end generation speeds.

Priority tier

Priority tier is for workloads that require higher reliability during peak traffic periods, at a higher price point. Priority tier is prioritized above Standard traffic and is less likely to be load shed (503 server overloaded). To use priority tier, set service_tier to "priority". Supported on OpenAI-compatible chat completions and on the Anthropic-compatible messages API:

curl https://api.fireworks.ai/inference/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $FIREWORKS_API_KEY" \
  -d '{
    "model": "accounts/fireworks/models/glm-5p3",
    "service_tier": "priority",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Priority tier is available on select models. Models and pricing are listed on the Serverless pricing page.

Fast

Fast is a high-speed mode, useful for interactive applications that require fast response speeds, at a higher price point. Fast variants aim for 100+ tokens per second of generated throughput. It is not a different model and the quality of the model remains the same. Fast is available for select models. To use Fast, change the model ID as listed below.

Model	`model` ID
Kimi K3 Fast	`accounts/fireworks/routers/kimi-k3-fast`
GLM 5.3 Fast	`accounts/fireworks/routers/glm-5p3-fast`

curl https://api.fireworks.ai/inference/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $FIREWORKS_API_KEY" \
  -d '{
    "model": "accounts/fireworks/routers/kimi-k3-fast",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Pricing is listed on the Serverless pricing page.

Ultra-fast (Preview)

Ultra-fast is a high-speed mode, useful for extreme latency sensitive applications that require fast response speeds, at a higher price point. Ultra-fast variants aim for 6-8x generation speed (up to 10x for agentic coding tasks). Like fast, it is not a different model and the quality of the model remains the same. Ultra-fast is in Preview and is available for the models listed below. Our preview modes are under active development and are subject to change quickly. To use Ultra-fast, change the model ID as listed below.

Model	`model` ID
Deepseek V4.1 Flash 1017	`accounts/fireworks/routers/deepseek-v4p1-flash-1017-ultrafast`

curl https://api.fireworks.ai/inference/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $FIREWORKS_API_KEY" \
  -d '{
    "model": "accounts/fireworks/routers/deepseek-v4p1-flash-1017-ultrafast",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Pricing is listed on the Serverless pricing page.

Serverless overview
US-only Serverless
Serverless quickstart
Text models
Anthropic compatibility — service_tier is supported on both OpenAI-compatible chat completions and the Anthropic messages API.

Get Started

Serverless Inference

Deployments

Models & Inference

Training

Managed Training

Training API

Fire Pass

Administration

Security & Compliance

Integrations

Reference

Priority tier

Fast

Ultra-fast (Preview)

​Priority tier

​Fast

​Ultra-fast (Preview)

​Related

Priority tier

Fast

Ultra-fast (Preview)

Related