ChatGPT

developed and maintained by and community

An AI chat widget for Zabbix dashboards.
Ask about your infrastructure without leaving the board - a server-proxied chat with OpenAI, Gemini, Claude or your own OpenAI-compatible endpoint, with admin-managed API keys that never reach the browser.

Features  ·  Examples  ·  Install  ·  FREE vs PRO  ·  Portal  ·  Docs

---

Why ChatGPT

Troubleshooting means switching between the dashboard and a chat window - and pasting company context into a personal AI account. ChatGPT puts the assistant on the dashboard itself: your admin provisions the API keys once (with per-team visibility), every request is proxied by the Zabbix server, and the raw key never travels to a browser.

What you can build

On-call assistant Ask "what does this trigger mean" or "how do I check disk latency" right next to the problem list.	Managed keys Admin-provisioned API tokens with visibility scopes - everyone / user groups / specific users.
Runbook drafting Turn a resolved incident into a first-draft runbook without leaving Zabbix.	Any provider  PRO OpenAI, Google Gemini, Anthropic Claude - or a self-hosted OpenAI-compatible endpoint (Ollama, vLLM).
Org-wide defaults  PRO Lock every widget to one company-approved token and model from Administration.	Tuning per token  PRO Temperature, top-p, max tokens and N - set once on the token, applied everywhere it is used.

Examples

Streamed answers - markdown-rendered, token by token	Live monitoring data - the model queries your Zabbix through MCP and says which tool it used
Attachments - drop in a log or a screenshot and ask about it	Your language - ask in any language, the answer comes back in it
Managed tokens - endpoint, MCP server and visibility per user group or user	Org-wide default - one connection for every widget on every dashboard

Configuration

The widget form is deliberately small - pick a provider, one of the admin-managed tokens and a model; everything sensitive (the key, a custom endpoint, tuning parameters) lives on the token under Administration → AI General. A saved key renders only as a mask of its length - the value never travels to the browser, and the reveal eye fetches it on an explicit click. PRO-only fields show inline (greyed with a padlock in the FREE edition).

For OpenAI, Gemini and Claude in PRO, the API-token profile shows a disabled Model field with a tooltip: select the model in each widget, or enable global configuration to override the widget choice. Custom profiles keep an editable Model field and Load models; the model configured in the selected Custom profile is used for that endpoint.

In PRO, switching providers and back while editing restores the model you selected for each provider, including when switching through a custom endpoint. The same applies when changing tokens between providers in global configuration. A saved model that is still supported stays selected when you reopen the form; an unavailable model falls back to the provider's default.

Install

Both FREE and PRO ship as GPG-signed deb / rpm packages from the initMAX repository - apt / dnf installs them and keeps them updated. Same flow for both editions; PRO just adds your personal repo token.

Easiest way - the guided installer on the Portal

Open the product page, pick your OS and edition, and copy the ready-made command. FREE is fully public (no login); PRO fills in your token once you sign in. There's a feedback box right there too.

→ Open the installer on the Portal

Prefer a plain archive? Every release also ships as a ZIP - FREE straight from the repo, PRO with your repo token - handy for offline or manual installs.

Then enable it in Administration → General → Modules, add an API token under Administration → AI General, and drop the widget on a dashboard. Done.

FREE vs PRO

Feature	FREE	PRO
AI chat on the dashboard (server-proxied, key never in browser)	✅	✅
Localised into all 27 Zabbix display languages	✅	✅
High availability ready	✅	✅
Admin-managed API tokens with visibility scopes	✅	✅
Streamed, markdown-rendered answers (sanitized)	✅	✅
Follow-latest scrolling with a return-to-latest control	✅	✅
OpenAI on the newest supported model	✅	✅
Send-by-Enter & streaming toggles	✅	✅
Upgrade-safe field schema (Zabbix 6.x → 7.x)	✅	✅
Provider choice - OpenAI · Gemini · Claude	❌	✅
Custom OpenAI-compatible endpoint (Ollama, vLLM, …)	❌	✅
Model selection from the curated per-provider list	❌	✅
Tuning parameters per token (temperature · top-p · max tokens · N)	❌	✅
Org-wide global configuration (one token for all widgets)	❌	✅
Live Zabbix data through MCP tools	❌	✅
Attachments (screenshots, logs and config files)	❌	✅
Remember conversation per user in the database	❌	✅
Stop generation & copy controls	❌	✅
Hide the initMAX logo	❌	✅
Licence	AGPLv3	Commercial

Requirements

Zabbix	6.0 · 6.2 · 6.4 · 7.0 · 7.2 · 7.4 - one package covers all
PHP	7.4 or newer
OS	Debian/Ubuntu · RHEL/Rocky/Alma/Oracle/Amazon · SUSE
Editions	FREE (public repo) · PRO (token-gated repo)
Languages	All 27 Zabbix display languages - the widget follows each user's own language setting
High availability	Ready. Conversations and admin tokens live in the Zabbix database, never on the node, so any frontend of an HA cluster can serve the next request

Support & links

• Documentation / Wiki
• Product page
• Portal - downloads, tokens, support tickets
• Source code (FREE, AGPLv3) - included in every package and published as a source archive on repo.initmax.com
• support@initmax.com

Supported AI models

Choose a provider in the product settings, then one of the models below. Custom accepts any OpenAI-compatible endpoint (you supply the model name)._

• OpenAI: GPT-6.1 Sol, GPT-6 Astra, GPT-6 Luna, GPT-6 Sol, GPT-5.6 Luna, GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.5, GPT-5.5 Pro, GPT-5.4, GPT-5.4 Mini, GPT-5.4 Nano, GPT-5.4 Pro
• Gemini: Gemini Flash (latest), Gemini Pro (latest), Gemini Flash Lite (latest), Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.5 Flash Lite, Gemini 3.1 Flash Lite, Gemini 3 Flash, Gemini 2.5 Pro
• Claude (Anthropic): Claude Opus 5.5, Claude Sonnet 5.5, Claude Fable 5.1, Claude Opus 5, Claude Sonnet 5, Claude Opus 4.8, Claude Sonnet 4.6, Claude Haiku 4.5
• Custom: any OpenAI-compatible endpoint

---

_{FREE: AGPLv3  ·  PRO: commercial  ·  © 2021-2026 initMAX s.r.o.}