Skip to main content
Allgate
Start free

Developers

01Developer experience

Keep your stack.
Change the endpoint.

From your first request to production: connect your SDK, choose a model and protocol, and keep usage and cost in view.

Create an API key Make a request
  1. 01Choose a model
  2. 02Create a key
  3. 03Make a request

Keep your SDK. Check parameters and capabilities per model.

allgate / quickstartExample request · not executed

A new destination for your requests

OpenAI-compatible base URLhttps://api.allgate.ai/v1
request.shHTTP · JSON
# export ALLGATE_API_KEY='YOUR_ALLGATE_API_KEY'curl --fail-with-body 'https://api.allgate.ai/v1/chat/completions' \  -H "Authorization: Bearer $ALLGATE_API_KEY" \  -H "Content-Type: application/json" \  --data-raw '{  "model": "gpt-5.5",  "messages": [    {      "role": "user",      "content": "Hello, Allgate!"    }  ]}'
Set ALLGATE_API_KEY on your server firstRead the API docs

API reference

Each protocol has its own request format.

Chat Completions/v1/chat/completionsResponses/v1/responsesAnthropic Messages/v1/messages

Example model: gpt-5.5Selected from available chat models. Check the catalogue for capabilities and pricing.

Your coding agent, through Allgate

Your terminal. Your workflow.

Configure the native API, model and API key for Claude Code or Codex in one setup flow.

API address for this agenthttps://api.allgate.ai/v1

Install a current agent CLI first. macOS / Linux need Bash 3.2+, curl and standard system tools; Windows needs PowerShell 5.1+. Codex configuration is verified with 0.163 and requires separate profile files and command-backed credentials.

1. Prepare your Allgate API key

Create a key in your account, then return here to configure your agent.

Create an Allgate key
Bash / 2. Configure the agent and keyRead the script
bash <(curl -fsSL https://allgate.ai/setup/allgate-setup.sh) --agent 'codex' --model 'gpt-5.5' --base-url 'https://api.allgate.ai'

After running the command, paste your key at “Allgate API key (hidden)” and press Enter.

3. Start the configured agentcodex --profile allgate

Codex uses a separate allgate.config.toml, preserving your default configuration and sign-in.

1. Prepare your Allgate API key

The setup command asks for your full key with hidden input, then saves and configures the agent credential automatically.

Back up before changing

Keep unrelated settings and permission rules. Matching configuration is left alone. Restore with the exact backup directory printed by setup.

You send the first request

Setup checks the public catalogue and configures local credentials. It makes no model call. Your first message is billed by actual usage.

02 / Choose. Connect. Verify.

Make the first request count.

Check availability and pricing units, create a key, then verify the result, usage and cost in your request logs.

  1. 01

    Choose an available model

    Find its current availability, exact model ID and pricing unit in the catalogue. The broader model library is a reference, not a list of callable models.

    Explore the catalogue
  2. 02

    Create a dedicated key

    Sign in, create an API key and set model allowlists, quota and RPM / TPM. Save the full key when it is created and keep it on your server.

    Manage API keys
  3. 03

    Verify the call and cost

    Get your configuration in the connection workspace. Send a first request, then check the model, status and cost in your logs. Real model calls are billed by usage.

    Open the connection workspace

Two variables. Server-side.

Export your Allgate key before running the example above. The Python and Node.js tabs include their install commands. Replace the placeholder key with your own.

.env / shell 
export ALLGATE_BASE_URL='https://api.allgate.ai/v1'
export ALLGATE_API_KEY='YOUR_ALLGATE_API_KEY'

Current example model: Gemini 4 Argongemini-4-argon View model details Open the full chat example

03 / Work with familiar tools

One gateway. Know the protocol.

Compatibility describes the request protocol. Tool calls, structured output, image input and individual parameters still depend on the model and upstream capabilities.

OpenAI

OpenAI-compatible endpoints

Use the official openai packages for Python and Node.js. The SDK base URL includes /v1. Authenticate with Authorization: Bearer and your Allgate key.

OpenAI SDK base URL
https://api.allgate.ai/v1
POST /v1/chat/completionsPOST /v1/responses
Chat reference Responses reference
Python SDK source Node.js SDK source
Anthropic

Anthropic Messages protocol

Send a Messages-shaped request to /v1/messages. Our HTTP examples use Authorization: Bearer. With the Anthropic SDK, set the API origin as the base URL; do not add /v1 twice.

Anthropic SDK base URL
https://api.allgate.ai
POST /v1/messagesAuthorization: Bearer $ALLGATE_API_KEY
Messages reference
Anthropic SDK source

Verify model, protocol and parameter support before moving production traffic.

04 / Language meets multimodal

Choose the capability. Find the endpoint.

These capabilities are covered by the gateway documentation. Counts reflect the current available catalogue; a documented endpoint does not guarantee model supply.

21 models available now

Language & reasoning

Chat, code, analysis and agents. Check context, tool support and output formats against the task you are building.

/v1/chat/completions
Read the reference Check the catalogue
2 models available now

Image generation

Turn a prompt into an image. Check which sizes, counts and response formats your chosen model actually supports.

/v1/images/generations
Read the reference Check the catalogue
3 models available now

Video generation

Create an asynchronous task, then poll its status. Read the result after success, and avoid creating the same task twice.

/v1/videos/generations
Read the reference Check the catalogue
No model currently listed as available

Audio & speech

Speech synthesis and transcription have API references. Confirm model supply, voices, formats and pricing units before integrating.

/v1/audio/speech /v1/audio/transcriptions
Read the reference Check the catalogue

Building a retrieval pipeline?

Embeddings and reranking use their own models and request formats.

Embeddings reference Rerank reference

05 / From a request to a product

Build the operational details in.

Protect access and credentials

Keep keys out of websites, mobile bundles and public repositories. Separate environments and apply model and quota limits. Revoke and rotate a key if it leaks.

Authentication & security

Consume the stream correctly

Request SSE with stream: true and consume events incrementally. Chat, Messages and Responses have different terminators. Inspect a failed stream before deciding to repeat it.

Streaming protocols

Recover with a clear boundary

On 429, respect Retry-After and use bounded backoff. Check credentials for 401, balance or quota for 402, and model access for 404. Avoid blindly replaying generation jobs.

Errors & rate limits

Keep a trace you can follow

Record x-allgate-request-id. Unary responses expose x-allgate-cost-micros; reconcile final usage and charges in console logs. Pricing units come from the catalogue.

View request logs

06 / Questions

A few details before you build.

Can I keep my existing OpenAI SDK?

For standard chat calls, keep the SDK and change the base URL, Allgate key and model ID. Validate your parameters, tools and stream handling first. Other endpoints follow their individual reference formats.

Is every model in the library callable?

The model library is a broader reference. The model catalogue is the source for callable models, availability and public pricing. Examples select an available chat model; your key allowlist may further restrict access.

How is a video task different from chat?

Video creation returns a task ID. Poll /v1/videos/generations/{id} with the same key, handle success or failure states, and set a reasonable polling interval and deadline.

Does SDK compatibility imply every model works?

Protocol compatibility, model availability and model capabilities are separate checks. Language, images, video, audio, embeddings and reranking also have different inputs, outputs and pricing units.

Keep the reference within reach.

Request shapes, streaming events, errors and limits. Validate the details for your workflow.

Open the API docs Explore pricing

Integration questions, answered.

How do I connect to the Allgate API?
Create an account and API key. For an OpenAI-compatible client, set the base URL to https://api.allgate.ai/v1 and choose a currently available model ID from the catalogue. Follow the protocol-specific setup in the reference for other SDKs. HTTP examples use a Bearer token; keep keys on your server.
Does a documented endpoint guarantee model availability?
The API reference describes request formats. The live catalogue lists models and their current routing availability. An endpoint does not mean that every model or capability has an available route.
Where can I check request formats and costs?
The public API reference covers authentication, request shapes, streaming and errors. Model detail pages publish billing units and base rates; your console shows the applicable account price.

Catalogue snapshot: · Public model data · Models and prices · Image and video Studio · API reference