This page maps what DEEIX Chat can do across the user console and the admin console. Use it to find the guide for a specific surface.
Where to find it#
- User console: sign in and use the left navigation for New chat, Recent, Files, Knowledge bases, and Skills & Prompts.
- Settings: open the user menu and choose Settings.
- Admin console: open the user menu and choose Admin.
Conversations and models#
| Capability | Details | Guide |
|---|---|---|
| Multimodal chat | Streaming replies, message branches, retry, edit, continue, fork, delete, like and dislike, and metadata such as model, tokens, latency, and cost. | New chat |
| Temporary chat | A chat that does not appear in history and is cleared when you leave the page. | New chat |
| Composer | File attachments, screenshots, voice input, message queue, / for prompts and skills, @ for files, tools, models, and knowledge bases. | New chat |
| Model kinds | Chat, audio, image generation, image editing, video generation, and video extension. | Models |
| Model parameters | Visual and JSON configuration, pass-through allowlists and denylists, and administrator-fixed defaults. | Capabilities & pass-through |
| Native tools | Provider-native tools for OpenAI, Claude, Google, and Grok. | Capabilities & pass-through |
| Protocols | OpenAI, Anthropic, Google/Gemini, xAI, OpenRouter, and custom OpenAI-compatible endpoints. | Upstreams |
| Routing | Platform models, upstream channels, route bindings, priority, weight, and automatic circuit breaking. | Models |
Files, knowledge bases, and retrieval#
| Capability | Details | Guide |
|---|---|---|
| File management | Upload, search, filter, sort, preview, extract, rename, download, and delete files, and check storage usage. | Files |
| Processing states | Waiting, queued, OCR running, extracting, vectorizing, ready, image ready, processing failed, index failed, no text, and index update required. | Files |
| Full-context injection | Inject extracted text into the conversation when the file fits the configured limits. | Files & retrieval |
| RAG | Chunking, embeddings, semantic retrieval, a similarity threshold, and a token budget. | Files & retrieval (advanced) |
| Knowledge bases | Personal and built-in knowledge bases, platform sources, availability, and vectorization. | Files |
| File strategy | Per-conversation Auto, Full context, or RAG handling. | Chat |
Tools and rendering#
| Capability | Details | Guide |
|---|---|---|
| MCP tools | MCP servers over Streamable HTTP, tool sync, per-tool enablement, display order, and per-message limits. | Tools |
| Image routing | Route image attachments to an MCP tool before the main model call. | Tools |
| Tool execution | Timeout, retry count, concurrency, max LLM requests, and max tool calls per message. | Tools |
| Rendering | Markdown, code, KaTeX math, Mermaid diagrams, HTML and SVG artifacts in a sandbox, and downloadable tables and diagrams. | New chat |
| Interactive components | Built-in, platform, and personal components such as charts, tables, calculators, Gantt charts, and quizzes. | New chat |
| Visual layout | An optional HTML visual layout prompt that guides the model to use safe inline HTML. | New chat |
| Process trace | The planning, retrieval, file, compaction, skill, and upstream stages for each message. | New chat |
Context and memory#
| Capability | Details | Guide |
|---|---|---|
| Context management | Message window, token budget, automatic context compression, and a user opt-out. | Conversation |
| Conversation context | Project instructions and conversation preferences used as context. | Projects |
| AI preference memory | Long-term preferences saved from the interface or from a reply. | Chat |
| Semantic recall | Optional embedding of message history to recall relevant earlier content. | Files & retrieval (advanced) |
| RAG evidence | Retrieval sources and match summaries shown in the process trace. | New chat |
Billing and usage#
| Capability | Details | Guide |
|---|---|---|
| Billing modes | Self-use, subscription billing, and usage billing. | Billing |
| Pricing | Token, per-call, per-second, tiered, and cache write pricing, with time-of-day rates. | Billing & pricing |
| Payments | Stripe and EPay, chosen by the user at checkout. | Billing |
| Plans and balance | Subscriptions, the entitlement queue, account balance, top-ups, and redemption codes. | Subscription |
| Usage records | Usage logs and statistics by user, permission group, and model. | Logs |
Identity and security#
| Capability | Details | Guide |
|---|---|---|
| Login | Username and email login, email registration, email verification, and password reset. | Sign-in & auth |
| Two-factor authentication | TOTP with a QR code or manual key, recovery codes, and re-generation. | Account |
| Identity providers | OIDC/OAuth2 providers with binding, verification, and auto-linking rules. | Sign-in & auth |
| Sessions | Active session list with location, IP, and last update time; log out all devices. | Account |
| Human verification | Cloudflare Turnstile for email registration. | Sign-in & auth |
| Account data | Public ID, profile, timezone, and permanent account deletion. | Account |
| Encryption | AES-GCM for upstream API keys, SSO secrets, MCP tokens, sensitive settings, and TOTP secrets. | Configuration |
Administration and operations#
| Capability | Details | Guide |
|---|---|---|
| Accounts | Users, roles, status, subscriptions, balances, and security actions. | Accounts |
| Permission groups | Rate multiplier, manual members, and accessible models. | Accounts |
| Upstreams | Channels, base URLs, API key rotation, protocols, headers, timeouts, and circuit breaking. | Upstreams |
| Models | Platform models, sources, routing, capabilities, and native tools. | Models |
| Announcements | Post-sign-in notices with type, pin state, and Markdown content. | Announcements |
| Logs | Audit, usage, authentication, and system events. | Logs |
| Content moderation | Moderation service and input/output policies, superadmin only. | - |
| Clients | Browser app, installable desktop client, and a shared client core for future platforms. | Desktop App, Architecture |
| Operations | Docker deployment, SQLite or PostgreSQL, memory or Redis cache, local or S3 storage, GeoIP, OpenTelemetry, structured logs, and a version endpoint. | Configuration |