Zed

Checked against Zed's docs on 2026-10-11

Zed's agent panel connects as an OpenAI-compatible provider: one block in Zed's settings.json, with the key in CREEKSIDE_API_KEY.

Base URL: https://api.creeksidecompute.ai/v1
API: OpenAI Chat Completions
Model id: qwen/qwen3.8-27b, or qwen/qwen3.6-35b-a3b for speed, or meta/muse-glimmer-30b, which always reasons first
Context window: 131072 tokens, up to 65536 of them output
Key: Make one at creeksidecompute.ai/account

Set it up

Make a key. Sign in at creeksidecompute.ai/account with your email and make a key. Each request is paid from prepaid credit: top up by card from $10, or redeem a promotional code.
Put the key in an environment variable. Add this line to your shell profile (for example ~/.zshrc), then restart Zed:
shell
```
export CREEKSIDE_API_KEY=<your key>
```

Add the provider to Zed's settings file, ~/.config/zed/settings.json:

settings.json

{
  "language_models": {
    "openai_compatible": {
      "Creekside": {
        "api_url": "https://api.creeksidecompute.ai/v1",
        "available_models": [
          {
            "name": "qwen/qwen3.8-27b",
            "display_name": "Qwen3.8 27B",
            "max_tokens": 131072,
            "max_output_tokens": 32768,
            "reasoning_effort": "medium",
            "capabilities": {
              "tools": true,
              "images": false,
              "parallel_tool_calls": false,
              "prompt_cache_key": false
            }
          },
          {
            "name": "qwen/qwen3.6-35b-a3b",
            "display_name": "Qwen3.6 35B A3B",
            "max_tokens": 131072,
            "max_output_tokens": 32768,
            "reasoning_effort": "medium",
            "capabilities": {
              "tools": true,
              "images": false,
              "parallel_tool_calls": false,
              "prompt_cache_key": false
            }
          },
          {
            "name": "meta/muse-glimmer-30b",
            "display_name": "Muse Glimmer 30B",
            "max_tokens": 131072,
            "max_output_tokens": 32768,
            "reasoning_effort": "medium",
            "capabilities": {
              "tools": true,
              "images": true,
              "parallel_tool_calls": false,
              "prompt_cache_key": false
            }
          }
        ]
      }
    }
  },
  "agent": {
    "default_model": { "provider": "Creekside", "model": "qwen/qwen3.8-27b" }
  }
}

Zed reads the key from an environment variable named after the provider: Creekside becomes CREEKSIDE_API_KEY. You can instead paste the key in Zed's settings, under AI, LLM Providers; Zed keeps it in the system keychain. Never put the key in settings.json. Give all four capabilities or none of them.

Pick the model in the agent panel's model selector, under Creekside. The default_model line makes it the model for new threads.

Thinking

With a reasoning_effort in the model's entry, the agent panel shows a thinking toggle. Off, Zed sends none and the model answers directly; on, it sends the level you chose, and the model thinks first. Without reasoning_effort there is no toggle and thinking stays off. Zed's Minimal level also counts as off here. Thinking tokens are billed as output.

Muse Glimmer 30B always reasons before it answers, so its toggle sets how long, not whether: off sends none, so it reasons briefly, at low, and on sends the level in its entry.

What does not work

Edit predictions. They do not use this provider. The agent panel, the inline assistant and commit messages do.
Images, on the Qwen models. They are listed for text only, which is what their entries' "images": false tells Zed. Muse Glimmer 30B reads images, so its entry says true.
Model discovery. Zed does not read /v1/models; it lists only the models in available_models.

Troubleshooting

401 invalid API key: The key is wrong, or Zed did not get it. Set CREEKSIDE_API_KEY and restart Zed, or paste the key again in Zed's settings, under AI, LLM Providers. A variable that is set wins over a pasted key. To test the key alone, curl -s https://api.creeksidecompute.ai/v1/models -H "Authorization: Bearer $CREEKSIDE_API_KEY" returns the model list when the key is good.
404 model … is not served: The model id is wrong. Use qwen/qwen3.8-27b exactly, in lower case, with the qwen/ prefix. GET /v1/models lists every id we serve. The name in available_models is the id Zed sends.
400 maximum context length is 131072 tokens: The conversation is longer than the model's window. Keep max_tokens at 131072 so Zed knows the window, then start a new thread.
402 or 429: A 402 means your prepaid credit does not cover the request: add credit at creeksidecompute.ai/account. A 429 means the key's rate, or the number of requests your account may run at once, is used up; agents that work in parallel reach that sooner. Wait and retry. The API reference lists the limits.

Other tools

Sources, read on 2026-10-11: OpenAI-compatible providers, API keys, agent panel.