Guides

Go to https://chat.waymore.ai/register and create your account using email, Google, or GitHub.

Check your inbox for a verification email and click the link to activate your account. If you signed up with Google or GitHub, this step is automatic.

After signing in, you will see the chat interface. Type a message in the input box and press Enter. The AI will respond in real-time with streaming text.

Try attaching a file to your message, creating a new chat session from the sidebar, or enabling two-factor authentication in your profile settings.

Navigate to your API Keys page in the dashboard. Click "Create New Key" and give it a descriptive name.

Important: Copy the API key immediately. It will only be shown once.

Use cURL or any HTTP client to send a chat completion request:

The response follows the OpenAI-compatible format. The assistant's message is in choices[0].message.content:

The usage field in the response shows token consumption. Monitor your overall usage from the dashboard or via the /api/usage/summary endpoint.

The following are fully compatible — no code changes needed for these:

• Request format: messages, model, stream, temperature
• Response format: choices, usage, finish_reason
• Streaming via Server-Sent Events (SSE)
• Function calling / Tools (tools, tool_choice, parallel tool calls)
• Message roles: system, user, assistant, tool
• Bearer token authentication

Replace the OpenAI base URL with the LLM Portal endpoint:

If you use the OpenAI Python or Node.js SDK, override the base_url and api_key:

Replace OpenAI model names (gpt-4o, gpt-4-turbo, gpt-3.5-turbo) with Waymore-A1-Instruct-1011. Use the /api/chat/models endpoint to list all available models.

The only changes required are the URL, auth header, and model name. Your existing request body works as-is:

The API auto-detects the Anthropic format and responds in the same style. All of the following Claude features work without any conversion:

• Top-level system field for system prompts
• Anthropic tool format with input_schema
• Tool use responses with tool_use content blocks
• Tool results via tool_result content blocks
• Anthropic response format (content[], stop_reason, input_tokens/output_tokens)

You can use Anthropic-style tool definitions directly. The API detects the format and returns tool calls in Anthropic style:

Send tool results back using the Anthropic tool_result content block:

If you use the Anthropic Python SDK, override the base_url and use your LLM Portal API key:

Set stream: true in your request body:

The response is a Server-Sent Events (SSE) stream. Each event contains a JSON chunk with the next token:

Each delta.content field contains a text fragment. Concatenate all fragments to build the complete response. The stream ends with data: [DONE].

Create a tool definition describing your function's name, purpose, and parameters using JSON Schema:

Tip: Write clear descriptions for both the function and each parameter. This helps the model decide when and how to use the tool.

Include the tools array and set tool_choice in your completion request:

When the model wants to use a tool, the response has finish_reason: "tool_calls" and includes a tool_calls array:

Parse the arguments field (it's a JSON string) and execute your function with those parameters.

Run your function using the arguments the model provided. In this example, you would call your weather API with location: "Athens, Greece" and get back the current conditions.

Send the function result back to the model in a follow-up request. Include the full conversation history plus a tool role message with the matching tool_call_id:

The model uses the tool result to generate a natural language answer:

Create a named session to group related messages:

Include the session_id in your completion request to associate it with the session:

List all messages in a session to review the conversation history:

Rename a session with PATCH or delete it entirely. Deleting a session removes all its messages.

Create separate API keys for development, staging, and production. This limits the blast radius if a key is compromised and makes it easy to revoke access for a single environment.

Only grant the permissions each key actually needs. A key used only for chat completions should not have image or vision permissions:

For production keys, restrict access to your server's IP addresses:

Configure rate limits to prevent runaway costs and set expiration dates for temporary access. Use rpmLimit for requests per minute, dailyLimit and monthlyLimit for token caps, and expiresInDays for auto-expiration.

Regenerate API key secrets periodically. The regenerate endpoint gives you a new secret while keeping the same configuration:

Note: The old key secret is immediately invalidated. Update your applications before or immediately after regenerating.

Regularly review each key's usage to detect anomalies:

Upload the file using the attachments endpoint. This returns a file URL you can reference in a message:

Include the attachment metadata when creating a message:

Each file can be up to 50MB. You can attach up to 5 files per message. Supported formats include images (PNG, JPG, GIF, WebP), documents (PDF, DOC, DOCX, TXT), data files (CSV, JSON, XML), and code files.

Save images, code snippets, notes, or any valuable content from your conversations:

Group related items into a collection with a custom name and color:

Add content items to your new collection:

Use powerful search and filtering to find content:

Query the usage summary endpoint for an overview of your consumption:

Narrow down usage data using query parameters:

The usage summary includes a cost breakdown by token type (input vs output) and by model. Use this to understand where your budget is going and optimize your prompts for cost efficiency.

Monitor individual API keys to track which integrations consume the most tokens: