Setup guide - From Qwen and Zed docs
Use Qwen in Zed
Zed documents OpenAI-compatible (and Anthropic-compatible) providers under language_models in settings.json or through Agent Settings > Add Provider. Model Studio documents both its OpenAI-compatible and Anthropic-compatible base URLs plus Qwen model ids.
Last verified 3 steps4 official sources
Steps
- Step 1.
Open Agent Settings (agent::OpenSettings), find Add Provider under LLM Providers, and fill in the provider name, API URL, model ID and context window. Or add the provider to settings.json (replace {WorkspaceId}):
JSON{ "language_models": { "openai_compatible": { "model-studio": { "api_url": "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1", "available_models": [ { "name": "qwen3.7-plus", "display_name": "Qwen3.7 Plus", "max_tokens": 1000000 } ] } } } } - Step 2.
Enter the API key in the provider settings UI, or set the generated environment variable: the provider ID in upper snake case plus _API_KEY (MODEL_STUDIO_API_KEY for provider id model-studio). Do not put API keys in settings.json.
Terminalexport MODEL_STUDIO_API_KEY="YOUR_API_KEY" - Step 3.
Coding Plan alternative: set api_url to the Coding Plan OpenAI endpoint and use the Coding Plan key.
JSON"api_url": "https://coding-intl.dashscope.aliyuncs.com/v1"
Caveats
- Combined setup: Zed's documented openai_compatible mechanism plus Model Studio's documented base URLs and model ids. There is no Zed-specific Model Studio guide.
- max_tokens 1000000 for qwen3.7-plus comes from the context_window in Model Studio's Codex model catalog.
- OpenAI-compatible models default to chat_completions true, tools true and images false in Zed; set capabilities.images to true for vision models.
- Zed's anthropic_compatible block is a documented alternative using the Model Studio /apps/anthropic endpoint.
- This setting covers the Zed Agent and Zed-owned AI features only, not External Agents or Terminal Threads.
More setup guides
Same tool, other providers
Related reading
- Run Qwen3.8-Flash-Next on a Gaming PC With StrataRun Qwen3.8-Flash-Next locally with Strata on a 12 GB GPU: hardware needs, speeds by quant, the license catch, and when the hosted API is the better call.
- Terminal-Universe Turns Agent Traces Into Training EnvironmentsTerminal-Universe, Qwen's paper on rebuilding terminal-agent trajectories into reusable environments, and why agent teams should keep replayable workspaces.
- Qwen3.8-27B vs Opus 4.6 Max: Local Agentic Coding BenchmarksQwen3.8-27B is a 27B dense Apache-2.0 model that beats Opus 4.6 Max on SWE-bench Pro (61.7 vs 53.4). Benchmarks, hardware, cost and when to run it.
- Qwen 3.8 Max Ships: 2.4T MoE, 1M Context, $2/$6 per MTok, Open Weights Next WeekQwen 3.8 Max is a 2.4T-parameter MoE with 95B active, a 1M context window and $2/$6 per million tokens on QwenCloud, with open weights promised next week.
- Qwen-UI-Agent Points at the Next GUI Agent RuntimeQwen-UI-Agent, Alibaba's GUI agent report, reads best as a product spec: mobile, desktop, browser, CLI and DeepSearch combined in one stateful agent runtime.
- $500 RL Fine-Tune of a 9B Open Model Beat GPT-5.6 Sol and Claude Opus 4.8 on Catalog ReviewA $500 GRPO fine-tune of Qwen 3.5 9B beat GPT-5.6 Sol and Opus 4.8 on catalog review, hitting 97% accuracy at a fraction of the cost per listing.
FAQ
How do I use Qwen in Zed?
Zed documents OpenAI-compatible (and Anthropic-compatible) providers under language_models in settings.json or through Agent Settings > Add Provider. Model Studio documents both its OpenAI-compatible and Anthropic-compatible base URLs plus Qwen model ids.
Which Qwen model ids work in Zed?
The official docs read for this guide list qwen3.7-plus, qwen3.6-plus and qwen3.7-max. Requests go to https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1.
Is this an official Qwen and Zed setup?
The steps draw on both Qwen's and Zed's docs. It was last checked against 4 official pages on October 17, 2026.
Sources
Last verified against 4 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.