Connect blindGram to OpenAI (ChatGPT)

Learn how to connect blindGram Assistant to official OpenAI models (GPT-4o, GPT-4o-mini, and Whisper). Follow step-by-step instructions to create an API key, configure voice and vision models, and resolve billing issues.

OpenAI provides powerful multimodal vision models and dedicated cloud speech recognition with Whisper, delivering highly accurate image descriptions and voice message summaries directly inside blindGram.

Prerequisites

• An OpenAI developer account at platform.openai.com.
• A prepaid credit balance (OpenAI developer API is pay-as-you-go and separate from a ChatGPT Plus web subscription).
• blindGram and blindGram Assistant installed on your Android device.

Step 1: Create an API key in OpenAI

1. Go to OpenAI Platform and sign in.

2. Ensure your account has prepaid credits under Settings > Billing. A small balance ($5) lasts a very long time for personal Telegram use.

3. Open the left sidebar and select API keys.

4. Tap Create new secret key.

5. Assign a name (for example, `blindGram Assistant`), keep default permissions, and tap Create secret key.

6. Tap Copy to save the key (starts with `sk-proj-...`).

> Important: OpenAI displays the secret key only once upon creation. Paste it directly into blindGram.

Step 2: Configure blindGram Assistant

1. Open blindGram on your Android device.

2. Go to Settings > blindGram Area > Assistant > AI.

3. Under The service:

• Tap Provider and select OpenAI.
• Tap API key and paste the key copied from OpenAI.
• Tap Save to confirm.

Step 3: Select models for text, vision, and speech

Because OpenAI supports both chat and audio transcription, you can configure both:

• Model (Text & Vision): #
• `gpt-4o-mini` (Recommended): Fast, economical, and delivers detailed image descriptions and summaries.
• `gpt-4o`: Maximum reasoning power for complex documents, math, and detailed scene analysis.
• Speech model (Audio transcription): #
• Select `whisper-1`. When present, voice summaries upload the voice recording directly to Whisper for near-perfect speech transcription across dozens of languages.

Step 4: Enable AI features

Turn on the features you want active:

• Describe image: In Describe image, enable Detailed description to send photos to GPT-4o for spoken scene analysis. #
• Voice summary: In Voice summary, enable Summarize with AI. A Summarize voice button will appear on voice messages. #
• Text summary: In Text summary, enable Summarize messages to condense long texts. #
• Ask AI: In Ask about messages, enable Ask AI about any message to ask follow-up questions about messages. #

Troubleshooting Common Problems

Problem: "You exceeded your current quota" or HTTP 429

• Cause: Your OpenAI account has no remaining prepaid credits, or your card has expired. #
• Solution: Note that ChatGPT Plus ($20/month web subscription) does not cover API usage. Go to platform.openai.com/billing, check your balance, and add $5 in prepaid credits. #

Problem: "Incorrect API key provided"

• Cause: The copied string is truncated or has extra spaces. #
• Solution: Create a new key in the OpenAI dashboard, tap the copy button, and paste it cleanly into the blindGram text field. #

Problem: Voice summaries failing on poor connections

• Cause: Voice files are sent to OpenAI Whisper over the network. #
• Solution: If your internet connection is slow or unstable, switch off the remote speech model row to let blindGram fallback to your phone's offline speech recognition engine. #

Frequently Asked Questions

Does my ChatGPT Plus subscription work here?

No. OpenAI separates the web ChatGPT subscription from the developer API. blindGram connects to the developer API, which is billed per token and typically costs pennies for daily personal use.

Where are my conversation transcripts saved?

All generated descriptions, summaries, and Q&A answers are stored locally on your device for one week under AI Conversations.