One API for agents that remember, act and show their work.

Claude, GPT, Gemini, Grok and open models behind one OpenAI-compatible endpoint, with agent sessions, memory and a trace for every call. Pay per token, with no subscription.

Start building — $5 freeSign in
POST /v1/chat/completionsclaude-sonnet-4-5
user

assistant

waiting for the first token

input
—
output
—
cost
$0.00000

Supported model families: Claude, GPT, Gemini, Grok, Qwen, Llama.

Everything an agent needs, behind one key.

Start with a single chat completion. Grow into long-running agents without changing providers, SDKs or billing.

Agents and sessions

Agents keep their state across turns and tool calls. Resume a session by id instead of replaying the whole transcript.

agentmodelstoolsmemorytracessession

Inference API

Every frontier model on one OpenAI-compatible endpoint. Switch models by changing a string.

Memory

Facts your agents learn persist across sessions, scoped to your organization.

Observability

Every request leaves a trace: prompt, tools, tokens, latency and cost to the micro-dollar.

Model families on one endpoint
6
Free credit when you sign up
$5
Scopes you can grant a key
5
Monthly fee
$0

Already using the OpenAI SDK? Change two lines.

Point the base URL at AdamAGI and use your key. Streaming, tool calls and usage come back in the shape your code expects.

curl https://api.adamagi.com/v1/chat/completions \
  -H "Authorization: Bearer adam_sk_live_…" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-5",
    "messages": [{ "role": "user", "content": "Plan my week" }],
    "stream": true
  }'

From sign-up to first token in a few minutes.

  1. Step 1: Create your account

    Sign up with email, Google or GitHub. $5 of credit lands in your workspace straight away.

  2. Step 2: Create an API key

    Pick its scopes, the models it may call and a spending limit. Rotate or revoke it at any time.

  3. Step 3: Ship

    Send requests from any OpenAI SDK. Watch usage, cost and traces update in the console as they happen.

Pay for tokens, not seats.

Prepaid credits, drawn down per request at the model's price. No subscription, no minimum, and every new workspace starts with $5.

Pay as you go

$5 free to start

  • Every model on one key
  • Usage and cost per key, per model, per day
  • Spending limits and rate limits per key
  • Organizations with roles for your team
Start building — $5 free
Price per million tokens, in US dollars
ModelInputOutput
Claude Haikuclaude-haiku-*$1.00$5.00
Claude Opusclaude-opus-*$15.00$75.00
Claude Sonnetclaude-sonnet-*$3.00$15.00
Geminigemini-*$1.25$10.00
GPT-5gpt-5*$1.25$10.00
Grokgrok-*$3.00$15.00
Other models$3.00$15.00

Questions, answered.

  • Any model on the price list, at the listed per-token price. For a mid-size model that is a few million tokens, plenty to build and test an agent end to end.

Your first agent is a key away.

Sign up, get $5 of credit and send your first request in minutes.

Start building — $5 free