Model & API platforms
Groq API 定價
免費方案適合技術團隊Groq 依使用量計價，而非依月計價；每個選項的費率列於下方表格中。 已於 2026年7月5日 對照官方定價頁面查核。
Extremely fast model inference API for latency-sensitive AI features.
Groq runs open models on its own hardware at very high speed, which matters for voice agents and anything users wait on. The price list is simple and linear; the catch is that you are limited to the open models Groq hosts.
方案與價格
| 方案 | 價格 | 你獲得什麼 |
|---|---|---|
| On-demand (Tokens-as-a-Service)我們的選擇 | 依用量計費Kimi K2 (0905): $1.00/$3.00. GPT-OSS 120B: $0.15/$0.60. GPT-OSS 20B & Safeguard 20B: $0.075/$0.30. Llama 3.3 70B: $0.59/$0.79. Llama 3.1 8B: $0.05/$0.08. | Free API key to start; Batch API 50% off; cached input 50% off
|
選擇此等級：On-demand (Tokens-as-a-Service)
On-demand is the only self-serve option and needs no commitment. Use the Batch API for work that does not need an instant answer.
表現良好之處
- Very fast inference for latency-sensitive features
- Simple per-token pricing with no subscription
- Batch API and cached input discounts
- Speech recognition and text-to-speech models too
需要留意之處
- Only the open models Groq chooses to host
- No frontier closed models
- Built for developers; no consumer product
以下情況可略過
- You need a specific closed frontier model
- Speed is not a problem for your use case
- You are not building with an API
最適合
Improve coding / developmentReduce software costs
Groq 定價相關問題
- Groq 在 2026 年要多少錢？
- Groq 依使用量計價，而非依月計價；每個選項的費率列於下方表格中。 已於 2026年7月5日 對照官方定價頁面查核。
- Does Groq have a free tier?
- You can start with a free API key; usage is then billed per token with no subscription or minimum commitment.
- What is Groq good for?
- Latency-sensitive features such as voice agents and live chat, where fast responses matter more than using the largest model.
Dapols team 於 2026年9月11日 審查。 回報價格變動
我們對此工具的證據
- 目錄狀態
- 推薦 — 可作為方案中的主要選擇
- 價格最後查核
- 2026年7月5日
- 證據可信度
- 中 — 已驗證，但供應商經常更動方案
分批重新查核。我們為每次查核標註日期，且絕不會因為排名而收取費用。 我們如何排名工具
Groq 適合你的企業嗎？
回答幾個問題，看看 Groq 是否符合你的工具組合與預算。