Skip to content

Fireworks AI

AI Infrastructure

Verification stateActiveFields checked (see sources):

Launched

Pricing

Pay-as-you-go

About

Inference platform for open and self-hosted models: per-token serverless inference, model ensembling and managed fine-tuning; the vendor claims up to 4x the throughput of providers wrapping vLLM.

InferenceOpen ModelsFine-tuning
Visit site→

Verification and sources

Verification state
Active
Fields checked (see sources)
Facts last changed
Version / scope
Serverless inference, on-demand deployments and managed fine-tuning; models listed on the page include gpt-oss-20b, Nemotron 3 Ultra, Kimi K3 and DeepSeek V4.1 Flash.
Runs on
Not recorded
Price checked

Not recorded

  • Per-token billing: the page lists Kimi K3 at $3/M input and $15/M output tokens; embeddings cost up to $0.016/M input tokens.

Price checked: Not recorded

Suits
Teams routing work across several open models and paying per finished task; the vendor sells inference, ensembling and training as one platform.
Limits and caveats
Billing is per token against pre-paid credits. Prices vary by model and serving path (Standard/Priority/Fast); the page publishes example unit prices rather than a flat monthly tier.
Sources

“Suits” and “Limits and caveats” are the desk’s assessment of the sources listed here, not a hands-on test.

Full disclaimer

Related coverage

We have no directly related coverage of this tool yet. Browse every tool in AI Infrastructure

Similar tools

硅基流动 (SiliconFlow)

An AI model inference cloud offering one-click deployment and low-cost inference for mainstream models like Qwen, DeepSeek and Llama.

Hugging Face

The world's largest open-source AI community, founded in 2016, hosting millions of models and datasets — the hub of open AI.

Ollama

An open-source local LLM runner released July 2023 — run Llama, Qwen, DeepSeek and more on your machine with one command.

LM Studio

A desktop app for local LLMs launched in 2023 — download and run open models with a GUI, zero-code local AI.