JetBrains AI Assistant on an OpenAI-compatible endpoint
Add it under Providers & API keys
As of 2026-10-07, AI Assistant's OpenAI-compatible provider works on QCode for GPT and GLM / Kimi / DeepSeek / Qwen only: in Settings | Tools | AI Assistant | Providers & API keys set URL to https://api.qcode.cc/openai/v1, paste a QCode key, then Test Connection. AI Assistant's Anthropic provider has no URL field, so it cannot reach QCode; for Claude install the Claude Code [Beta] plugin with ANTHROPIC_BASE_URL https://api.qcode.cc/api.
Updated 2026-10-07
Point this tool at the QCode endpoint — one key for Claude, GPT and Chinese models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Takeaways
OpenAI-compatible fields
Official settings page: Provider, API Key, URL, and Tool calling (Tool calling is OpenAI-compatible only). Then Test Connection.
Docs fetched
jetbrains.com/help/ai-assistant/use-custom-models.html and settings-reference-providers-and-api-keys.html (the old settings-reference-models.html now redirects there), re-checked 2026-10-07. The settings page title is still Providers & API keys.
How far the URL goes
For QCode use https://api.qcode.cc/openai/v1, for GPT and GLM / Kimi / DeepSeek / Qwen only. Do not paste /chat/completions.
What this page covers
Third-party models for chat and related features. AI Completion is a separate block; official docs require FIM, which typical chat models do not provide.
What third-party models mean in AI Assistant
AI Assistant defaults to cloud models through the JetBrains AI service. Official docs also let you bring an API key for Anthropic, Google, OpenAI, or an OpenAI-compatible endpoint (examples: llama.cpp, LiteLLM). QCode's OpenAI-compatible endpoint uses that last option and serves GPT and GLM / Kimi / DeepSeek / Qwen only; Claude in JetBrains goes through the Claude Code plugin. This is not a client-id hack.
What people still search
In 2026-08 threads people still ask how JetBrains BYOK works. @JetBrainsHelp on 2026-08-21: a personal ChatGPT subscription cannot connect directly to AI Assistant Chat; use an OpenAI API key via BYOK. This page answers the official OpenAI-compatible fill-in, then what to do after you paste QCode.
Timeline
Use third-party and local models (28 July 2026) and the Providers & API keys settings page (15 July 2026): Settings | Tools | AI Assistant | Providers & API keys. Older help text may still say Models & API keys; trust the title in the IDE.
Official list includes OpenAI-compatible endpoints. That provider needs a URL, API Key, and a Tool calling toggle. The 64 000-token default context window is documented for local models, not as the remote-gateway default on this page.
@JetBrainsHelp: a personal ChatGPT subscription cannot attach directly to AI Assistant Chat; use an API key via BYOK. Docs also say IDE Services / JetBrains Central admins may lock third-party providers.
Confirmed vs watch-outs
Confirmed
AI Assistant documents OpenAI-compatible: URL + API Key + Test Connection. Provider models show up in the chat model selector. Docs re-checked 2026-10-07.
Watch-outs
Not every AI Assistant feature works with third-party models; official docs suggest activating JetBrains AI alongside. AI Completion is independent of the chat provider and needs FIM. IDE Junie Agent in YouTrack LLM-22660 (2026-01) then required first-party Anthropic or OpenAI — this page does not treat Junie as the same path. No header-override or proxy recipes.
Which path
AI Assistant · OpenAI-compatible (this page)
Use this when you have a custom base URL. For QCode enter https://api.qcode.cc/openai/v1 for GPT and GLM / Kimi / DeepSeek / Qwen. Assign remote models to Core features / Instant helpers under Model Assignment.
First-party providers / Junie
Anthropic, Google, and OpenAI first-party keys use their own Provider rows. Junie is a different product — do not copy this page into Junie. A ChatGPT personal subscription is not an API key.
What to do next
① Open Settings | Tools | AI Assistant | Providers & API keys and choose OpenAI-compatible under Third-party AI providers. ② URL https://api.qcode.cc/openai/v1, paste the QCode API key, set Tool calling if needed, Test Connection, Apply. ③ In AI Chat open the model selector and pick a GPT or GLM / Kimi / DeepSeek / Qwen model from that provider's section (ids on /models, e.g. gpt-6-sol, glm-5.3); claude-* is not available on this endpoint. For inline completion read the AI Completion block; do not assume a chat model does FIM.
On QCode
QCode's OpenAI-compatible endpoint is https://api.qcode.cc/openai/v1 and serves GPT and GLM / Kimi / DeepSeek / Qwen only; claude-* returns model_not_available_on_endpoint there. For Claude in JetBrains, install the Claude Code CLI and the Claude Code [Beta] plugin (the plugin reuses the CLI's configuration), then set ANTHROPIC_BASE_URL to https://api.qcode.cc/api, with the same key as ANTHROPIC_AUTH_TOKEN. Which model ids work is the live /models list — this page does not freeze a catalog as always-on-sale.
FAQ
Can AI Assistant talk to a custom OpenAI-compatible gateway?
Yes. Use third-party and local models lists OpenAI-compatible endpoints. The settings page has URL and Test Connection.
What is the settings page called now?
The title fetched 2026-08-30 is Providers & API keys, path Settings | Tools | AI Assistant | Providers & API keys. Older help may still say Models & API keys; trust the IDE.
How far down should the URL go?
Stop at /openai/v1, i.e. https://api.qcode.cc/openai/v1. Do not append /chat/completions.
Does ChatGPT Plus count as an API key?
@JetBrainsHelp on 2026-08-21: a personal ChatGPT subscription cannot connect directly to AI Assistant Chat. Use an OpenAI API key via BYOK. A QCode key is that API-key path.
Can Junie take the same URL?
This page is AI Assistant only. YouTrack LLM-22660 (2026-01) then said Junie Agent needed first-party Anthropic or OpenAI. Junie CLI has a separate custom-model JSON path. Do not treat this page as a Junie recipe.
Will inline completion work after this?
Not necessarily. Official docs split AI Completion into its own block and require FIM for inline completion. Typical chat models do not offer that. Third-party models can also leave some features unavailable.
Sources
JetBrains Use third-party and local models (jetbrains.com/help/ai-assistant/use-custom-models.html, footer 28 July 2026) and Providers & API keys (settings-reference-providers-and-api-keys.html, 15 July 2026), fetched 2026-08-30, re-checked 2026-10-07. @JetBrainsHelp 2026-08-21 on ChatGPT subscriptions. YouTrack LLM-22660 cited as an observation only. QCode endpoint values follow the docs.qcode.cc JetBrains and endpoints pages (fetched 2026-10-07).
Add OpenAI-compatible in AI Assistant
Official URL field + QCode /openai/v1 for GPT and GLM; Claude goes through the Claude Code plugin.
Related
Cline on OpenAI-compatible APIs
Custom base URL in a VS Code extension.
Claude Desktop third-party API guide
Configure a Gateway in developer mode: Anthropic Messages API only, Claude models only, no ANTHROPIC_BASE_URL, plus the Claude Code differences.
Claude Code VS Code Extension: API Key and Custom Gateway
Credentials in claudeCode.environmentVariables, settings.json shared with the CLI, plus login prompt and model selection fixes.
Not affiliated with JetBrains. Behavior follows JetBrains help. This page is not a guide to bypassing any vendor subscription auth.