KeyoAPI
Free AI API — $0 Models, No Credit Card
KeyoAPI publishes a free AI API catalog: fixed $0 model IDs, no credit card to start, fair-use limits. Register a key and set model to any ID in the table below.
A free API key, no credit card
Create an account, mint a key, point OPENAI_BASE_URL to https://www.keyoapi.xyz/v1, and call chat completions with a free model ID. That is the whole free API key path.
Free models ($0)
| Free model ID | Paid twin (if any) | Price | Family | Links |
|---|---|---|---|---|
Atria-dawn-v2 |
— | $0 | Other | Try |
DeepSeek-Prover-V2-7B |
— | $0 | DeepSeek | Try |
deepseek-v4-flash-free |
deepseek-v4-flash |
$0 | DeepSeek | Try |
deepseek-v4-pro-free |
deepseek-v4-pro |
$0 | DeepSeek | Try |
dots-3-note-preview:free |
— | $0 | Other | Try |
gemini-robotics-er-2-preview:free |
— | $0 | Try | |
gemma-4-26b:free |
— | $0 | Try | |
gemma-4-31b-it:free |
— | $0 | Try | |
glm-5.2-free |
glm-5.2 |
$0 | 智谱 | Try |
glm-5.2-thinking:free |
— | $0 | GLM | Try |
glm-5.2:free |
— | $0 | GLM | Try |
glm-5.3-flash-search:free |
— | $0 | GLM | Try |
glm-5.3-flash-think-search:free |
— | $0 | GLM | Try |
glm-5.3-flash-thinking:free |
— | $0 | GLM | Try |
glm-5.3-search:free |
— | $0 | GLM | Try |
glm-5.3-think-search:free |
— | $0 | GLM | Try |
glm-5.3-thinking:free |
— | $0 | GLM | Try |
glm-5.3:free |
— | $0 | GLM | Try |
kimi-k3-free |
kimi-k3 |
$0 | Moonshot | Try |
laguna-s-2.1:free |
— | $0 | Other | Try |
ling-3.0-flash-fin:free |
— | $0 | Ant Group | Try |
mistral-large-3-675b:free |
— | $0 | Mistral | Try |
muse-glimmer-30b:free |
— | $0 | Other | Try |
nemotron-3-super-120b-a12b:free |
— | $0 | NVIDIA | Try |
nemotron-3-ultra-550b-a55b:free |
— | $0 | NVIDIA | Try |
nemotron-3.5-lightning-30b-a3b:free |
— | $0 | NVIDIA | Try |
nemotron-3.5-lightning:free |
— | $0 | NVIDIA | Try |
qwen3.6-35b-a3b:free |
— | $0 | Qwen | Try |
qwen3.8-27b:free |
— | $0 | Qwen | Try |
step-3.7-flash:free |
— | $0 | StepFun | Try |
Call the free ID exactly as listed (including -free or :free suffixes). Upstream free pools and catalog membership can change — confirm live availability on /pricing.
Free by family
Browse by family in the table above (DeepSeek, GLM, Qwen, Google, NVIDIA, and more). The highest-traffic free search intent is still “free API key” / “free LLM API” — this hub is that landing.
Quick start (curl)
curl https://www.keyoapi.xyz/v1/chat/completions \
-H "Authorization: Bearer $KEYO_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4-flash-free",
"messages": [{"role":"user","content":"Hello"}]
}'
Free rules (read before you ship)
- Price: listed free IDs use fixed $0 ModelPrice — not a timed coupon that flips to paid overnight.
- Fair use: free traffic has rate / concurrency limits. Limits can tighten under abuse. Do not treat free IDs as an unlimited production SLA.
- Availability: some
:freeIDs depend on upstream free pools and may be added or removed. Check Model Square before hard-coding. - Catch (honest): you get real capability at $0; you do not get paid-tier priority or guaranteed throughput.
FAQ
Is it really free?
Yes for the $0 IDs listed above: no credit card required to start. Create an account, mint an API key, and set model to a free ID.
What's the catch?
Fair-use rate limiting and lower priority. Exact RPM can change; if you need predictable limits, use metered (paid) IDs.
Free vs paid?
Free IDs are $0 with fair-use limits. Paid IDs are token- or request-metered with higher priority — preferred for production.
Can I use it in production?
You can. For anything user-facing or latency-sensitive, we recommend metered IDs. Free is ideal for prototypes, demos, CI smoke tests, and eval harnesses.
Why one hub page?
One hub keeps free IDs, limits, and examples in sync. Building dozens of near-identical free model pages would drift and dilute quality.