API Keys & Providers
Conjure supports 13+ transcription and LLM providers through a bring-your-own-key (BYOK) model. You add your API keys, and Conjure calls the provider APIs directly from your machine. No data passes through Conjure's servers.
Why BYOK?
- No subscription required -- use free tiers or pay only for what you use
- Privacy -- your audio and text go directly to the provider you choose
- Flexibility -- mix and match providers for transcription and post-processing
- Control -- manage your own usage limits and billing
Provider Comparison
Transcription (Speech-to-Text)
| Provider | Cost | Speed | Quality | Notes |
|---|---|---|---|---|
| Local Whisper | Free | Moderate | Good | CPU or GPU, no internet needed |
| Groq | Free tier | Fast | Excellent | whisper-large-v3, 25K min/month free |
| Deepgram | $0.0043/min | Fast | Excellent | Nova-2 model, real-time streaming |
| AssemblyAI | $0.0065/min | Fast | Excellent | Real-time streaming support |
| OpenAI | $0.006/min | Moderate | Excellent | whisper-1 model |
| Azure | $0.01/hr | Moderate | Excellent | Azure Speech Services |
| ElevenLabs | $0.008/min | Fast | Good | Streaming support |
Post-Processing (LLM Text Generation)
| Provider | Cost per 1M tokens | Speed | Notes |
|---|---|---|---|
| Groq | Free tier | Very fast | llama-3, mixtral models |
| Gemini | Free tier | Fast | Google's models, generous limits |
| Ollama | Free (local) | Varies | Run any model locally |
| DeepSeek | $0.14-$0.28 | Fast | Very affordable, strong coding |
| OpenAI | $2.50-$10.00 | Moderate | GPT-4o, GPT-4o-mini |
| Claude | $3.00-$15.00 | Moderate | Anthropic, strong reasoning |
| OpenRouter | Varies | Varies | Access to 100+ models |
| Azure OpenAI | Same as OpenAI | Moderate | Enterprise Azure deployments |
Recommended Free Setup
If you want the best experience at zero cost:
| Function | Provider | Why |
|---|---|---|
| Transcription | Groq | Free tier includes 25K minutes/month of whisper-large-v3 |
| Post-processing | Groq | Same free key works for LLM text generation |
| Backup STT | Local Whisper | Works offline, no API needed |
This setup gives you cloud-quality transcription and AI post-processing without spending a cent.
Recommended free setup
Use Groq for both transcription and post-processing (free tier, no credit card required) with local whisper as an offline fallback.
Adding API Keys
- Open Settings (gear icon in the sidebar)
- Navigate to the API Keys section
- Click "Add API Key"
- Select the provider from the dropdown
- Paste your API key
- (Optional) Give the key a label for identification
- Click "Test" to verify the key works
- Save
You can add multiple keys for the same provider and multiple providers simultaneously.
Getting API Keys
Groq (recommended free option)
- Go to console.groq.com
- Create a free account
- Navigate to API Keys
- Click "Create API Key"
- Copy the key and add it to Conjure
OpenAI
- Go to platform.openai.com
- Sign in or create an account
- Navigate to API Keys in settings
- Click "Create new secret key"
- Copy the key and add it to Conjure
Gemini (Google)
- Go to aistudio.google.com
- Sign in with your Google account
- Click "Get API key" or navigate to API keys
- Create a key for a new or existing project
- Copy and add to Conjure
Ollama (local)
- Install Ollama from ollama.ai
- Pull a model:
ollama pull llama3 - In Conjure, add an Ollama provider -- no API key needed, just the local endpoint (
http://localhost:11434)
DeepSeek
- Go to platform.deepseek.com
- Create an account and add credits (minimum $2)
- Navigate to API Keys
- Create a key and add it to Conjure
Claude (Anthropic)
- Go to console.anthropic.com
- Create an account
- Navigate to API Keys
- Create a key and add it to Conjure
Usage Tracking
Conjure tracks API usage per key:
- Token counts for post-processing calls (input + output tokens)
- Cost estimates based on provider pricing
- Per-key badges in Settings showing total calls and estimated cost
- Home page card with aggregated usage across all keys
Usage data is stored locally in SQLite and never sent anywhere. Cost estimates are approximations based on published provider pricing.
Provider Selection
You can configure different providers for different functions:
- Transcription provider -- which API handles speech-to-text
- Post-processing provider -- which LLM cleans up your transcript
- Agent provider -- which LLM powers agent mode conversations
This lets you optimize for cost, speed, and quality independently. For example, use Groq for fast transcription and Claude for high-quality agent conversations.
Editing and Removing Keys
- Click on any API key in Settings to edit its label or update the key value
- Delete keys you no longer use -- transcription history retains usage stats even after key deletion
- Test keys after editing to ensure they still work
Keys stored locally
API keys are stored in your local SQLite database and never sent to Conjure servers. All API calls go directly from your machine to the provider.
Security
API keys are stored in your local SQLite database. They are:
- Never sent to Conjure servers -- all API calls go directly from your machine to the provider
- Included in data exports -- if you export your data, API keys are included (with a security warning)
- Not encrypted at rest -- they're stored as plaintext in SQLite, protected by your OS file permissions
WARNING
Treat your Conjure database file as sensitive. Anyone with access to conjure.db can read your API keys.