Orgo - Cloud computers for AI agents

Send a message to an AI model and have it drive an Orgo computer on your behalf. The model sees the screen, clicks, types, and runs shell commands until your request is satisfied, then returns the final assistant message. Use this endpoint as a drop-in replacement for openai.chat.completions.create.

The response is wire-compatible with OpenAI’s chat completions. Any SDK that points at https://www.orgo.ai/api/v1 works unchanged - the only required extension is passing a computer_id to bind the agent to a running VM.

Endpoint

POST https://www.orgo.ai/api/v1/chat/completions

Auth: Authorization: Bearer sk_live_...

Request

string

required

UUID of a running Orgo computer - returned as the id field by Create computer. The authenticated user must own the computer or be a member of its workspace.

array

required

Array of { role, content } objects. role must be user or assistant. The first message must have role user. When continuing a thread, only the new messages need to be passed - prior history is loaded from the thread.

string

default:"claude-sonnet-5"

Model identifier. One of claude-sonnet-5, claude-opus-5, claude-opus-4.8, claude-sonnet-4.6, or claude-opus-4.6. See Models below.

boolean

default:"false"

If true, responses are streamed as OpenAI-format Server-Sent Events. The connection stays open until the agent finishes or the request is cancelled.

string

Continue a previous multi-turn session. When set, the server loads the thread’s history, runs the agent with full context, and appends new messages on completion. If omitted, a new thread is created automatically and its ID returned in the response.

integer

default:"100"

Maximum number of agent steps (screenshot → action → observation) before the run stops. Increase for long, multi-phase workflows.

Headers

string

required

Bearer sk_live_... - your Orgo API key. Get one at orgo.ai/settings/api-keys.

string

Bring-your-own Anthropic key. When present, requests bill directly to your Anthropic account instead of drawing from your Orgo credit balance. Enterprise plans can omit this and still bypass metering.

Response

Non-streaming

string

Request identifier (chatcmpl-...). Also returned as the X-Request-Id response header.

string

Always chat.completion.

integer

Unix timestamp (seconds).

string

The model ID used for the request.

array

Single-element array containing the final assistant message.

Show choice

integer

Always 0.

object

{ role: "assistant", content: string } - the model’s final text response after the agent loop terminates.

string

Always stop on success.

object

Token counts for the full agent loop (all intermediate turns, not just the final message).

Show usage

integer

Total input tokens.

integer

Total output tokens.

integer

Sum of input and output tokens.

object

Orgo-specific metadata not in the OpenAI spec.

Show orgo

string

ID of the thread this completion belongs to. Pass it back in a later request to continue the conversation.

integer

Number of agent steps executed (screenshot + action cycles).

number

Credits consumed by this request, in cents. 0 for BYOK and enterprise plans.

integer

Remaining credit balance after settlement. Omitted for BYOK and enterprise plans.

Response headers

Header	Description
`X-Request-Id`	Echoes the `id` field. Include it when reporting issues.
`X-Thread-Id`	Thread ID. Present when a thread is created or continued.

Streaming

When stream: true, the server returns text/event-stream with standard OpenAI chat.completion.chunk events:

data: {"id":"chatcmpl-01HX...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]}

data: {"id":"chatcmpl-01HX...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"Opening Chrome"},"finish_reason":null}]}

data: {"id":"chatcmpl-01HX...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":1240,"completion_tokens":312,"total_tokens":1552}}

data: [DONE]

Only the model’s text output is streamed. Tool calls, screenshots, and intermediate reasoning happen server-side and are not exposed as deltas. The X-Thread-Id header is set on the initial response so clients can associate the stream with a thread before the first chunk arrives.

Examples

from openai import OpenAI

client = OpenAI(
    base_url="https://www.orgo.ai/api/v1",
    api_key="sk_live_a3bb189e8bf93888",
)

response = client.chat.completions.create(
    model="claude-sonnet-4.6",
    messages=[
        {"role": "user", "content": "Open Chrome and search for 'claude computer use'"}
    ],
    extra_body={"computer_id": "a3bb189e-8bf9-3888-9912-ace4e6543002"},
)

print(response.choices[0].message.content)
print(f"Thread: {response.orgo['thread_id']}")
print(f"Cost: ${response.orgo['cost_cents'] / 100:.4f}")

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://www.orgo.ai/api/v1",
  apiKey: "sk_live_a3bb189e8bf93888",
});

const response = await client.chat.completions.create({
  model: "claude-sonnet-4.6",
  messages: [
    { role: "user", content: "Open Chrome and search for 'claude computer use'" },
  ],
  // computer_id is an Orgo extension - cast to bypass SDK types
  computer_id: "a3bb189e-8bf9-3888-9912-ace4e6543002",
} as any);

console.log(response.choices[0].message.content);

curl https://www.orgo.ai/api/v1/chat/completions \
  -H "Authorization: Bearer sk_live_a3bb189e8bf93888" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4.6",
    "computer_id": "a3bb189e-8bf9-3888-9912-ace4e6543002",
    "messages": [
      {"role": "user", "content": "Open Chrome and search for claude computer use"}
    ]
  }'

Streaming with raw SSE

import json, httpx

with httpx.stream(
    "POST",
    "https://www.orgo.ai/api/v1/chat/completions",
    headers={"Authorization": "Bearer sk_live_a3bb189e8bf93888"},
    json={
        "model": "claude-sonnet-4.6",
        "computer_id": "a3bb189e-8bf9-3888-9912-ace4e6543002",
        "messages": [{"role": "user", "content": "Open github.com"}],
        "stream": True,
    },
    timeout=300,
) as r:
    for line in r.iter_lines():
        if not line.startswith("data: "):
            continue
        payload = line[6:]
        if payload == "[DONE]":
            break
        delta = json.loads(payload)["choices"][0].get("delta", {})
        if "content" in delta:
            print(delta["content"], end="", flush=True)

curl -N https://www.orgo.ai/api/v1/chat/completions \
  -H "Authorization: Bearer sk_live_a3bb189e8bf93888" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4.6",
    "computer_id": "a3bb189e-8bf9-3888-9912-ace4e6543002",
    "messages": [{"role": "user", "content": "Open github.com"}],
    "stream": true
  }'

Continuing a thread

Pass the thread_id returned by a previous response. The agent loads the prior conversation and picks up where it left off.

first = client.chat.completions.create(
    model="claude-sonnet-4.6",
    messages=[{"role": "user", "content": "Open Chrome and go to github.com"}],
    extra_body={"computer_id": computer_id},
)

follow_up = client.chat.completions.create(
    model="claude-sonnet-4.6",
    messages=[{"role": "user", "content": "Search for 'orgo'"}],
    extra_body={
        "computer_id": computer_id,
        "thread_id": first.orgo["thread_id"],
    },
)

Bring your own Anthropic key

Pass x-anthropic-key to bill the request against your own Anthropic account. Orgo credits are not consumed and orgo.cost_cents is 0.

curl https://www.orgo.ai/api/v1/chat/completions \
  -H "Authorization: Bearer sk_live_a3bb189e8bf93888" \
  -H "x-anthropic-key: sk-ant-api03-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-opus-4.6",
    "computer_id": "a3bb189e-8bf9-3888-9912-ace4e6543002",
    "messages": [{"role": "user", "content": "Take a screenshot"}]
  }'

Example response

{
  "id": "chatcmpl-01HX8W3Z6PN2Q1MRT7YV9K4BFA",
  "object": "chat.completion",
  "created": 1745136000,
  "model": "claude-sonnet-4.6",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Done. Chrome is open on google.com with the search results for 'claude computer use'."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 1240,
    "completion_tokens": 312,
    "total_tokens": 1552
  },
  "orgo": {
    "thread_id": "thr_01HX8W3Z6PN2Q1MRT7YV9K4BFA",
    "steps": 6,
    "cost_cents": 3.84,
    "credit_balance_cents": 9616
  }
}

Models

Model	Best for	Notes
`claude-sonnet-5`	Default. Fast and capable - handles most workflows.	Tool version `computer_20251124`.
`claude-opus-5`	The most capable model, for complex multi-step tasks.	Tool version `computer_20251124`. Most expensive per token.
`claude-opus-4.8`	Prior-generation Opus.	Tool version `computer_20251124`.
`claude-sonnet-4.6`	Prior-generation Sonnet.	Tool version `computer_20251124`.
`claude-opus-4.6`	Prior-generation Opus.	Tool version `computer_20251124`.

Orgo’s OpenAI-compatible endpoint uses dotted model IDs (claude-sonnet-4.6). If you’re calling Anthropic’s native SDK directly in the Claude Computer Use guide, use the hyphenated form (claude-sonnet-4-6) - that’s Anthropic’s canonical identifier.

Billing

Requests are metered after the agent loop finishes. For non-BYOK, non-enterprise users, Orgo places a credit hold at request start and settles against actual token usage on completion. If the request errors before any tokens are consumed, the hold is refunded. Pricing matches Anthropic’s token rates plus a small platform margin; exact per-request cost is returned in orgo.cost_cents.

Errors

Status	Code	When
400	`invalid_json`	Request body is not valid JSON.
400	`invalid_model`	`model` is not one of the supported identifiers (see Models).
400	`missing_computer_id`	`computer_id` was not supplied.
400	`empty_messages`	`messages` array is empty.
400	`invalid_message_order`	First message does not have role `user`.
400	`context_overflow`	Conversation exceeds the model’s context window. Start a new thread or prune history.
401	`invalid_api_key`	Missing or invalid `Authorization` header.
402	`credits_exhausted`	Credit balance too low. Add credits at orgo.ai/settings/billing. Includes `balance_cents`.
404	`computer_not_found`	Computer does not exist, is not accessible, or does not belong to your project.
500	`internal_error`	Unexpected server error. Retry with exponential backoff. Include the `X-Request-Id` header when reporting.

All error responses have the shape:

{
  "error": {
    "type": "invalid_request",
    "message": "computer_id is required. Pass the ID of a running Orgo computer.",
    "code": "missing_computer_id"
  }
}

Overview

Computer Use API

Workspaces

Computers

Computer Lifecycle

Computer Actions

Files

Templates

Create chat completion

Endpoint

Request

Headers

Response

Non-streaming

Response headers

Streaming

Examples

Streaming with raw SSE

Continuing a thread

Bring your own Anthropic key

Example response

Models

Billing

Errors

​Endpoint

​Request

​Headers

​Response

​Non-streaming

​Response headers

​Streaming

​Examples

​Streaming with raw SSE

​Continuing a thread

​Bring your own Anthropic key

​Example response

​Models

​Billing

​Errors

Endpoint

Request

Headers

Response

Non-streaming

Response headers

Streaming

Examples

Streaming with raw SSE

Continuing a thread

Bring your own Anthropic key

Example response

Models

Billing

Errors