Skip to main content
Watch: I Built an AI App Without Leaving Databricks

Setup guide - Documented by Qwen

Use Qwen in Codex CLI

Model Studio documents Codex CLI setup through a custom model_provider in ~/.codex/config.toml pointing at the OpenAI-compatible endpoint (/compatible-mode/v1), with OPENAI_API_KEY as env_key. Models that support the Responses API work with the latest Codex; Chat/Completions-only models (and the whole Coding Plan) need Codex 0.80.0 with wire_api = "chat".

Last verified 8 steps2 official sources

Steps

  1. Step 1.

    Install Codex (Node.js v18.0 or later) and verify.

    Terminal
    npm install -g @openai/codex
codex --version
  2. Step 2.

    Set OPENAI_API_KEY to your Model Studio API key (macOS zsh example; use ~/.bash_profile for bash).

    Terminal
    echo 'export OPENAI_API_KEY="YOUR_API_KEY"' >> ~/.zshrc
source ~/.zshrc
  3. Step 3.

    Pay-as-you-go, Responses API (for models such as qwen3.7-max): edit ~/.codex/config.toml. Region base URLs: Beijing https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1, Singapore https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1, US (Virginia) https://{WorkspaceId}.us-east-1.maas.aliyuncs.com/compatible-mode/v1.

    TOML
    model_provider = "Model_Studio"
model = "qwen3.7-max"
[model_providers.Model_Studio]
name = "Model_Studio"
base_url = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1"
env_key = "OPENAI_API_KEY"
wire_api = "responses"
  4. Step 4.

    Optional model switching: create ~/.codex/model-catalog.local.json with the model metadata from the Alibaba page, then point config.toml at it. Switch with codex -m (for example codex -m qwen3.8-max) or /model in the TUI.

    TOML
    model_catalog_json = "~/.codex/model-catalog.local.json"
  5. Step 5.

    Pay-as-you-go, Chat/Completions-only models: install Codex 0.80.0 and use wire_api = "chat".

    TOML
    npm install -g @openai/codex@0.80.0

model_provider = "Model_Studio"
model = "qwen3.6-plus"
[model_providers.Model_Studio]
name = "Model_Studio"
base_url = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1"
env_key = "OPENAI_API_KEY"
wire_api = "chat"
  6. Step 6.

    Coding Plan (Chat/Completions only, older Codex such as 0.80.0): set OPENAI_API_KEY to the Coding Plan key.

    TOML
    model_provider = "Model_Studio_Coding_Plan"
model = "qwen3.7-plus"
[model_providers.Model_Studio_Coding_Plan]
name = "Model_Studio_Coding_Plan"
base_url = "https://coding-intl.dashscope.aliyuncs.com/v1"
env_key = "OPENAI_API_KEY"
wire_api = "chat"
  7. Step 7.

    Token Plan Personal Edition, Responses API (latest Codex).

    TOML
    model_provider = "Model_Studio_Token_Plan_Personal"
model = "auto"
[model_providers.Model_Studio_Token_Plan_Personal]
name = "Model_Studio_Token_Plan_Personal"
base_url = "https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1"
env_key = "OPENAI_API_KEY"
wire_api = "responses"
  8. Step 8.

    Verify: open a new terminal and start Codex. If the chat interface launches, the configuration is correct.

    Terminal
    codex

Caveats

  • Newer Codex versions reject wire_api = "chat" with wire_api = "chat" is no longer supported or unknown configuration field wire_api; the Coding Plan therefore requires pinning an older Codex such as 0.80.0.
  • The pay-as-you-go model catalog on the Alibaba page lists qwen3.8-max, qwen3.7-max, deepseek-v4-pro and glm-5.2 for the Responses API path.
  • unexpected status 401 Unauthorized usually means a key from a different plan or region; unexpected status 404 Not Found means base_url or wire_api is wrong.
  • Third-party switchers such as CC-Switch may report "domestic models not supported" because their health-check probe is rejected with 400; Alibaba says this does not affect actual Codex use.
  • Replace {WorkspaceId} with your Workspace ID; keys are region-specific.

More setup guides

Related reading

FAQ

How do I use Qwen in Codex CLI?

Model Studio documents Codex CLI setup through a custom model_provider in ~/.codex/config.toml pointing at the OpenAI-compatible endpoint (/compatible-mode/v1), with OPENAI_API_KEY as env_key. Models that support the Responses API work with the latest Codex; Chat/Completions-only models (and the whole Coding Plan) need Codex 0.80.0 with wire_api = "chat".

Which Qwen model ids work in Codex CLI?

The official docs read for this guide list qwen3.7-max, qwen3.8-max, qwen3.6-plus, qwen3.7-plus, qwen3.8-flash, qwen3.6-flash and auto. Requests go to https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1.

Is this an official Qwen and Codex CLI setup?

Qwen documents this setup in its own docs. It was last checked against 2 official pages on October 17, 2026.

Sources

Last verified against 2 official pages.

Newsletter

New models, set up the day they ship

One email per week. New model releases, the coding tools that support them and how to wire them up.