Raven Pro vs Open Source

Raven comes in two flavors:

Raven Pro - Download the app, sign in, and start using it immediately. Everything is managed for you.
Raven Open Source - Clone the repo, build from source, and bring your own API keys.

Both versions share the same core - the same stealth overlay, echo cancellation, transcription, AI, modes, and sessions. Pro adds managed infrastructure, cloud sync, and additional features on top.

Feature Comparison

Feature	Raven Pro	Open Source
Setup	Download, install, sign in	Clone repo, install dependencies, build from source
API keys	Managed - nothing to configure	You provide your own (Deepgram + Anthropic/OpenAI)
Transcription	Deepgram + AssemblyAI with auto-fallback	Deepgram (your key)
AI	Managed Claude (backend proxy)	Claude or GPT (your key)
Choose model in Settings	No — Fast/Deep toggle on overlay only	Yes — Haiku/Sonnet/Opus or GPT-5.x
Meeting detection	Auto-detects Zoom, Teams, Meet (Apple Silicon)	—
Post-meeting insights	Summaries, action items, topics, sentiment	—
Fast/Deep toggle	Yes (Haiku 4.5 / Sonnet 4.6)	—
Cloud sync	Sessions, modes, documents across devices	—
Sign in	Google or email	—
Stealth overlay	Yes	Yes
Echo cancellation	Yes	Yes
Modes & quick actions	Yes + cloud sync	Yes (local)
Document upload	Yes + cloud sync	Yes (local)
Incognito mode	Yes	Yes
Session history & search	Yes + cloud sync	Yes (local)
20+ languages	Yes	Yes
Auto-updates	Yes	—

Raven Pro Plans

Raven Pro has two tiers:

	Free	Pro
Price	$0	$29.99/mo or$ 24.99/mo (yearly)
AI requests/day	5	Unlimited
Session duration	2 minutes	Unlimited
Post-meeting insights	—	Yes
Cloud sync	Yes	Yes
Meeting detection	Yes	Yes

All limits on the Free tier reset at midnight UTC each day. See Plans & Billing for details on upgrading.

Open Source - Bring Your Own Keys

The open-source version requires you to provide API keys during the in-app onboarding:

Deepgram - for real-time transcription (free tier available at console.deepgram.com)
Anthropic or OpenAI - for AI assistance

In Settings → API Keys, pick your provider and model (e.g. Claude Opus 4.7 for maximum reasoning, or Haiku / GPT-5.4 Mini for speed). All requests go directly from your machine to these services. Your keys are stored locally and never sent to Raven’s servers. There are no usage limits beyond what your API provider imposes. To get started with the open-source version, see Build from Source.

Language Support

Both versions support transcription in 20+ languages:


English	Japanese	Bengali
Hindi	Korean	Dutch
Spanish	Chinese	Polish
French	Arabic	Russian
German	Tamil	Thai
Italian	Turkish	Ukrainian
Portuguese	Vietnamese

Multilingual mode auto-detects the spoken language. The output language (what language the AI responds in) can be set independently: English, Auto (matches transcript), Hindi, Spanish, French, German, Italian, Portuguese, Japanese, Korean, or Chinese. Both settings are in Settings > Language.

Privacy

Raven Pro — sessions, modes, and documents can sync to the cloud for cross-device access. Anonymous crash reporting (Sentry) runs by default with PII stripped. Product analytics (PostHog) are opt-in. See the Privacy Policy.
Open source — nothing is sent to Raven’s servers except direct API calls to Deepgram and your chosen AI provider. No Raven cloud sync, no PostHog.
Both — incognito mode skips saving session content locally when enabled.

Full terms: Terms of Service · Privacy Policy

​Feature Comparison

​Raven Pro Plans

​Open Source - Bring Your Own Keys

​Language Support

​Privacy

Feature Comparison

Raven Pro Plans

Open Source - Bring Your Own Keys

Language Support

Privacy