Every model.One key.
GPT, Claude, Gemini, Kimi, DeepSeek and more through one fast API. No email. No KYC. No subscription. Generate a private account key and start building.
Your stack.
Just smarter.
Keep your SDK, tools, and workflow. Change the base URL, paste your SafeAPI key, and choose any model in the catalog.
Generate your account
No email form or password. Your long, checksummed account key is your entire login.
Create an API key
Use separate revocable keys for Oh My Pi, production apps, and experiments.
Make your first request
Model discovery gives Oh My Pi the exact IDs automatically. Usage appears in your dashboard in real time.
# Install Oh My Pi, then create ~/.omp/agent/models.yml
providers:
safeapi:
baseUrl: https://safeapi.cc/v1
apiKey: SAFEAPI_API_KEY
api: openai-responses
discovery:
type: openai-models-list
export SAFEAPI_API_KEY='sk-safe_...'
omp models refresh safeapi
omp -p --model safeapi/gpt-6.2-astra "Ship it"Less friction.
More tokens.
The gateway watches capacity, shares it fairly, and gets out of the way. You see exactly what every request cost and how every model performs.
Capacity that scales sideways.
Healthy upstream keys share load by real in-flight demand. Cooldowns and pre-stream failover keep a bad route from becoming your bad day.
Your agents get a fair lane.
Per-user concurrency and fair scheduling stop one busy account from taking the whole pool. No mysterious premium tier.
TPS without the marketing math.
Generation speed excludes queue time and first-token latency. Aggregates are token-weighted, not an average of averages.
No identity form. No password.
A 256-bit account key replaces your profile. Prompts and completions are never written to application logs.
Pick your brain.
Normal names for humans. Exact IDs for your code. Straightforward PAYG pricing per million tokens.
Full model catalog| Model | Exact API ID | Input / 1M | Output / 1M | Avg TPS / 24h | Capabilities |
|---|---|---|---|---|---|
| GPT-6.2 AstraOpenAI | gpt-6.2-astra | $0.0672 | $0.252 | Collecting | reasoningcodeflagship |
| GPT-6.2 SolOpenAI | gpt-6.2-sol | $0.0672 | $0.252 | Collecting | reasoningcodegeneral |
| GPT-6.1 SolOpenAI | gpt-6.1-sol | $0.0672 | $0.252 | Collecting | generalcode |
| GPT-6.1 TerraOpenAI | gpt-6.1-terra | $0.0538 | $0.242 | Collecting | fastcode |
| GPT-6 AstraOpenAI | gpt-6-astra | $0.202 | $0.756 | Collecting | reasoningflagship |
| GPT-6 LunaOpenAI | gpt-6-luna | $0.0054 | $0.0202 | 50.3 | fastcode |
Before you ask.
Do I need an email or identity verification?
No. Signup generates a long account key. That key is your only login credential, so save it somewhere safe. There is no email reset flow.
How do crypto top-ups work?
Choose an amount and currency to create a unique invoice. Your balance updates automatically when the payment reaches the required status. Network fees and confirmation times depend on the asset you choose.
Can I use Oh My Pi, Cline, Codex, or Claude Code?
Yes. Oh My Pi is our recommended CLI and supports SafeAPI model discovery. OpenAI Chat Completions, Responses, and Anthropic Messages routes cover the other popular coding agents.
What does average TPS actually mean?
For streaming requests, SafeAPI divides output tokens by generation time after the first token. The 24-hour model value is weighted by actual tokens and generation time. Non-streaming measurements are labeled separately.
Do you cache my prompts?
SafeAPI does not create a shared semantic response cache. Supported provider-side prompt caching fields pass through normally, and eligible cached tokens are billed using the published cache rates. Prompts and responses are not logged.