Reproduction steps
- Obtain a Digital Ocean model access key
- Configure LLM Providers -> Add Provider -> Anthropic
- Configure Digital Ocean, with the API URL
https://inference.do-ai.run/v1/messages per the docs
- Set the model name to
anthropic-claude-4.6-sonnet, Max Output Tokens 64000, Max Tokens 200000
- Save Provider
- Start a new thread
- Make sure to select the newly added model specifically (separate bug)
- Send a prompt
Current vs. Expected behavior
Current
Expected
I can run Digital Ocean's Messages API sample successfully after substituting my DO model access key, so I expect Zed to be able to use the same endpoint.
Zed version and system specs
Zed Dev 1.8.0
Commit: 5e514f4
Attach Zed log file
Zed.log
INFO [agent::thread] Thread::send called with model: anthropic-claude-4.6-sonnet
ERROR [agent::thread] Turn execution failed: HTTP response error from Digital Ocean's API: status 404 Not Found - "<!DOCTYPE html>\n<html>\n<head>\n <title>DigitalOcean - Maintenance</title>\n <link href=\"https://assets.digitalocean.com/public/favicon.png\" rel=\"shortcut icon\" type=\"image/png\">\n <style type=\"text/css\">\n body {\n background-color: #27a1e3;\n margin: 0px;\n padding: 0px;\n }\n div.header{\n background-color: #E2F4FE;\n background-image: url('https://assets.digitalocean.com/public/waves.png');\n background-repeat: repeat-x;\n background-position: center bottom;\n padding-top: 150px;\n }\n div.main{\n overflow: auto;\n width: 960px;\n margin: auto;\n margin-top: 50px;\n }\n div.text{\n width: 450px;\n float: left;\n }\n div.text p{\n color: #fff;\n font-size: 18pt;\n font-family: helvetica;\n }\n div.mascot{\n width: 450px;\n float: right;\n }\n div.copyright{\n clear: both;\n }\n div.copyright p{\n color: #fff;\n text-align: center;\n font-size: 10pt;\n font-family: helvetica;\n padding-top: 60px;\n }\n </style>\n</head>\n\n<body>\n<div class=\"header\">\n</div>\n<div class=\"main\">\n <div class=\"text\">\n <a href=\"https://www.digitalocean.com\"><img src=\"https://assets.digitalocean.com/public/logo.svg\"></a>\n <p>DigitalOcean is currently under maintenance and will be back shortly!</p>\n </div>\n <div class=\"mascot\">\n <img src=\"https://assets.digitalocean.com/public/mascot.png\" />\n </div>\n</div>\n<div class=\"copyright\">\n <p>DigitalOcean<sup>TM</sup>, Inc. All Rights Reserved.</p>\n</div>\n</body>\n</html>\n"
Relevant Zed settings
settings.json
"language_models": {
"anthropic_compatible": {
"Digital Ocean": {
"api_url": "https://inference.do-ai.run/v1/messages",
"available_models": [
{
"name": "anthropic-claude-4.6-sonnet",
"max_tokens": 200000,
"max_output_tokens": 64000,
"extra_beta_headers": [],
"capabilities": {
"tools": true,
"images": false,
"prompt_caching": false
}
}
]
}
}
}Relevant Keymap
No response
(for AI issues) Model provider details
Anthropic via Digital Ocean
If you are using WSL on Windows, what flavor of Linux are you using?
None
Reproduction steps
https://inference.do-ai.run/v1/messagesper the docsanthropic-claude-4.6-sonnet, Max Output Tokens64000, Max Tokens200000Current vs. Expected behavior
Current
Expected
I can run Digital Ocean's Messages API sample successfully after substituting my DO model access key, so I expect Zed to be able to use the same endpoint.
Zed version and system specs
Zed Dev 1.8.0
Commit: 5e514f4
Attach Zed log file
Zed.log
Relevant Zed settings
settings.json
Relevant Keymap
No response
(for AI issues) Model provider details
Anthropic via Digital Ocean
If you are using WSL on Windows, what flavor of Linux are you using?
None