AI email writing and reply generation using context from received emailsWorks in any text field across any website via browser extensionSupport for 20+ languagesGrammar and tone correctionContext-aware response generationOne-click reply drafting in GmailCustom instructions for personalized writing style
Features
Multiple model sizes (7B, 13B, 70B, 405B) for flexible deploymentInstruction-tuned variants optimized for chat and reasoningStrong performance on coding, mathematics, and long-form text generationSupport for up to 131K context window in Llama 3.1Available for commercial use under Meta licenseSeamless integration with Hugging Face, vLLM, Ollama, and major frameworks
Pricing Tiers
Pricing Tiers
Free$0/mo/monthly
Limited AI generations per week
Premium/monthly
Unlimited generations, priority access
Pricing Tiers—
Integrations
Integrations
GmailGoogle DocsLinkedInTwitter/XFacebookAny web-based text field (via Chrome extension)
Integrations
Hugging Face TransformersvLLMOllamaLangChainAmazon BedrockAzure AITogether AIReplicate
Platforms
Platforms
chrome-extension
Platforms
webapidesktop-macdesktop-windows
Pros
Pros
Works across any website text field, not limited to a single platform
Deep Gmail integration reads email context to generate accurate replies
Supports 20+ languages for multilingual users
No need to leave the current tab — drafts appear inline
Free tier available for light use without a credit card
Pros
Exceptionally permissive open-source license enabling commercial use without restrictions
Wide parameter range (7B to 405B) supports use cases from edge devices to data centers
Competitive or superior reasoning and coding performance versus larger proprietary models
Massive community, extensive documentation, and countless pre-built integrations and tools
Can be run on-premises or any cloud with full data control and no API costs
Cons
Cons
Limited to Chrome (and Chromium-based) browsers — no Firefox or Safari support
No standalone web app or mobile app; entirely extension-dependent
Free tier has generation limits that can be restrictive for daily heavy use
Less suitable for long-form content like blog posts or reports
Output quality depends heavily on the prompt/context provided by the user
Cons
Requires infrastructure expertise to self-host; no built-in SaaS platform from Meta
Inference costs and latency unpredictable if self-hosting without proper optimization
Smaller context windows (up to 131K tokens) than some competitors like Claude
Less specialized optimization for specific domains; requires fine-tuning for expert performance