Everything the work
actually needs

Built around the full arc of how serious work gets done — from raw research to collaborative delivery. No feature bloat, just the tools that matter.

Multi-model routing

@mention any model in any message. Route one prompt to many models and compare outputs side by side in real time.

Torqon context engine

Optimizes memory, summaries, recent conversation, and token budgets before inference so models receive cleaner context.

Focus modes

Switch between Code, Research, Writing, Analysis, Think, and General — each tuned with a precision prompt layer for optimal results.

Rich file support

Upload images, PDFs, CSVs, and code files. Project-level asset libraries make files available across every session.

Everyone workflow

Multiple models collaborate on a single thread, building on each other's responses until the best answer surfaces.

Web search mode

Toggle live web search when prompts need current information. Meridian retrieves, cites, and synthesizes in context.

Image generation

Dedicated image models available directly in the workspace. Generate, refine, and export without switching tools.

Encrypted key storage

Bring your own keys for OpenRouter, OpenAI, Anthropic and custom providers — all encrypted at rest.

Usage analytics

Token counts, image generation, and cost estimates by model and project. Know exactly what your work costs.

Stripe payments

Top-up store with geo-aware pricing fallback. Upgrade, manage limits, and pay seamlessly through Stripe.

Security, auth,
and accounts

Full user account system with email/password, Google OAuth, Apple OAuth, demo login, password reset, and session validation — all included from day one.

01

Multi-provider auth

Email/password, Google OAuth, Apple OAuth, and demo login. Users choose how they sign in.

02

Session management

Automatic session validation, refresh tokens, and secure logout across all devices.

03

Settings & personalization

Default model, thought trace, streaming behavior, security preferences, and account controls.

04

Artifact & code preview

Generated code appears in a side workspace panel. Edit, iterate, and export without losing context.

Type / to do
anything.

Every action in Meridian is one keystroke away. Type / in any chat input to trigger commands — no menus, no clicks.

CommandWhat it does
/helpOpen slash command list modal
/shortcuts  /keys  /keyboardOpen keyboard shortcuts modal
/new  /clearStart a fresh chat
/stopAbort all active streams
/regen  /retryRegenerate the last response
/model <name>Fuzzy-match and pin a model
/everyone  /allFan-out to all workspace models
/effort <low|med|high|max>Set reasoning effort level
/thinkToggle extended thinking mode
/webToggle live web search
/imageToggle image generation tool
/humanToggle humanise tool
/translate <lang>Set target language for translation
/bullets  /numbered  /table  /codeSet output format type
/attach  /upload  /fileOpen file picker
/compress  /summarizeCompress current context
/upgrade  /pro  /planOpen upgrade modal

See every capability
in action

Start with the free tier. Every capability is available — only limits and model access differ by plan.