Skip to main content
Connect Glyph to OpenAI’s API to use GPT-4, GPT-4o, GPT-4 Turbo, and other models.

​
Prerequisites

  • OpenAI API account: platform.openai.com
  • API key with appropriate permissions
  • Sufficient API credits

​
Setup

1

Get API Key

  1. Log in to OpenAI Platform
  2. Navigate to API Keys in your account settings
  3. Click Create new secret key
  4. Copy the key (starts with sk-)
Store your API key securely. OpenAI only shows it once.
2

Open Glyph AI Settings

Go to Settings → AI and select the OpenAI profile.
3

Add API Key

  1. Click Set API Key in the authentication section
  2. Paste your OpenAI API key
  3. Click Save
The key is stored in .glyph/app/ai_secrets.json in your space directory.
4

Select Model

Click the Model dropdown. Glyph fetches available models from OpenAI’s API.Popular models:
  • gpt-4o - Latest GPT-4 Omni (recommended)
  • gpt-4o-mini - Faster, more affordable GPT-4
  • gpt-4-turbo - GPT-4 Turbo with vision
  • gpt-4 - Original GPT-4
  • gpt-3.5-turbo - Fast and cost-effective
5

Test Connection

Open the AI panel and send a test message. You should receive a response from your selected model.

​
Configuration

​
Provider Settings

  • Service: openai
  • Base URL: https://api.openai.com/v1 (default)
  • Authentication: Bearer token (API key)

​
Custom Endpoint

To use a custom OpenAI endpoint (proxy, Azure OpenAI, etc.):
  1. Set Base URL to your endpoint
  2. Add any required headers in Custom Headers
  3. Enable Allow Private Hosts if using localhost
Base URL: https://<resource-name>.openai.azure.com/openai/deployments/<deployment-name>
Headers:
[
  { "key": "api-key", "value": "your-azure-api-key" },
  { "key": "api-version", "value": "2024-02-15-preview" }
]

​
Model Selection

Glyph fetches the latest model list from OpenAI’s /v1/models endpoint.
ModelUse CaseContext Window
gpt-4oGeneral purpose, multimodal128K tokens
gpt-4o-miniFast, affordable128K tokens
gpt-4-turboAdvanced reasoning128K tokens
gpt-4Original GPT-48K tokens
gpt-3.5-turboSimple tasks, speed16K tokens
Glyph displays models returned by the API. If a model isn’t listed, type its ID manually in the model field.

​
Chat Completion Models Only

Glyph uses the /v1/chat/completions endpoint. Ensure your selected model supports chat completions.
Models like text-davinci-003 or gpt-3.5-turbo-instruct are not chat models. If you select one, you’ll see:
Model 'gpt-3.5-turbo-instruct' is not chat-completions compatible.
Select a chat model (e.g., gpt-4o, gpt-4-turbo, gpt-4o-mini).

​
Features

​
Chat Mode

Conversational interaction without tools:
  • Back-and-forth dialogue
  • Faster responses (no tool overhead)
  • Best for brainstorming and discussion

​
Create Mode

AI with workspace access:
  • File reading via read_file tool
  • Search notes with search_notes tool
  • List files with list_dir tool
  • Best for research and knowledge retrieval

​
Context Attachment

Attach files or folders to ground responses:
  • Attach via context menu in AI panel
  • Mention files with @filename syntax
  • Context sent in system message
  • Token estimates shown before sending

​
API Usage and Billing

Glyph makes direct API calls to OpenAI:

​
Cost Estimation

Use the context manifest to estimate costs:
  1. Attach context in AI panel
  2. View token estimate in manifest
  3. Calculate cost using OpenAI pricing
Example:
  • 10K input tokens + 1K output tokens with gpt-4o
  • Input: 10,000 × 0.0025/1K=0.0025 / 1K = 0.025
  • Output: 1,000 × 0.010/1K=0.010 / 1K = 0.010
  • Total: ~$0.035 per request

​
Rate Limits

OpenAI enforces rate limits based on your usage tier:
  • Free tier: 3 requests/min, 200 requests/day
  • Tier 1+: Higher limits based on usage history
If you hit rate limits, Glyph displays the error from OpenAI. Wait before retrying or upgrade your tier.

​
Troubleshooting

​
”API key not set for this profile”

Solution: Add your OpenAI API key in Settings → AI.

​
”model list failed (401)”

Solution: Your API key is invalid or expired. Generate a new key from OpenAI Platform.

​
”model list failed (429)”

Solution: You’ve hit OpenAI’s rate limit. Wait before retrying.

​
”This model is not chat-completions compatible”

Solution: Select a chat model like gpt-4o, gpt-4-turbo, or gpt-4o-mini.

​
Model list is empty

Solution: Type the model ID manually (e.g., gpt-4o). The model will work even if the list fetch failed.

​
Responses are slow

Possible causes:
  • Large context (10K+ tokens)
  • Complex tool usage in create mode
  • OpenAI API latency
Solution: Try a faster model like gpt-4o-mini or reduce context size.

​
Security Best Practices

  • Never commit .glyph/app/ai_secrets.json to version control
  • Rotate API keys if exposed
  • Use separate keys for different projects
  • Set spending limits in OpenAI dashboard

​
Next Steps

Chat Modes

Learn about chat vs create modes

Context Management

Attach notes to conversations

OpenRouter

Access 100+ models via OpenRouter

Profiles

Manage multiple AI profiles