continuedev/continue · error · Error

`env.apiVersion` is a required configuration property for Az

Error message

`env.apiVersion` is a required configuration property for Azure OpenAI

What it means

For Azure OpenAI's native API (apiType 'azure-openai'), every request URL must include the apiVersion query parameter, which Azure uses to select the API surface. The adapter validates this in _getAzureBaseURL and throws before sending a request that Azure would reject with a 400 ('api-version query parameter is required').

Source

Thrown at packages/openai-adapters/src/apis/Azure.ts:62

    // Copy search params to separate object for OpenAI
    const queryParams: Record<string, string> = {};
    for (const [key, value] of url.searchParams.entries()) {
      queryParams[key] = value;
    }

    url.pathname = url.pathname.replace(/\/$/, ""); // Remove trailing slash if present
    url.search = ""; // Clear original search params

    // Default is `azure-openai` in docs, but previously was `azure`
    if (this._isAzureOpenAI(config.env?.apiType)) {
      if (!config.env?.deployment) {
        throw new Error(
          "`env.deployment` is a required configuration property for Azure OpenAI",
        );
      }

      if (!config.env?.apiVersion) {
        throw new Error(
          "`env.apiVersion` is a required configuration property for Azure OpenAI",
        );
      }

      const basePathname = `openai/deployments/${config.env.deployment}`;

      url.pathname =
        url.pathname === "/" ? basePathname : `${url.pathname}/${basePathname}`;

      queryParams["api-version"] = config.env.apiVersion;
    }

    return {
      baseURL: url.toString(),
      defaultQuery: queryParams,
    };
  }

View on GitHub (pinned to 5522c6f44c)

Solutions

1. Add env.apiVersion with a valid Azure OpenAI GA version, e.g. '2024-02-01'
2. Use the latest preview/GA version matching your resource's region and features if the GA version errors
3. If on the azure-api-key gateway mode, switch apiType so apiVersion is not required

Example fix

// before
{ provider: 'azure', apiType: 'azure-openai', apiKey: '...', deployment: 'my-deploy' }

// after
{ provider: 'azure', apiType: 'azure-openai', apiKey: '...', deployment: 'my-deploy', apiVersion: '2024-02-01' }

const AZURE_API_VERSION = '2024-02-01';
if (config.env?.apiType === 'azure-openai' && !config.env?.apiVersion) {
  config.env.apiVersion = AZURE_API_VERSION; // safe default
}

const isCompleteAzureConfig = (c: AzureConfig): boolean =>
  c.env?.apiType === 'azure-api-key' || Boolean(c.env?.deployment && c.env?.apiVersion);

try { new AzureApi(config); } catch (e) {
  if (e instanceof Error && e.message.includes('env.apiVersion')) {
    config.env.apiVersion = '2024-02-01';
    return new AzureApi(config);
  }
  throw e;
}

When it happens

Trigger: Configuring apiType: 'azure-openai' with a deployment but no env.apiVersion; the defaultQuery getter calls _getAzureBaseURL which enforces the requirement.

Common situations: Porting code from the openai npm SDK where apiVersion is passed per-request rather than in config; forgetting that Azure requires the dated version string (e.g. '2024-02-01'); config files trimmed down to apiKey only.

Understand the failure class

Background: "X is required", "must be set", "cannot be empty": the missing-required-config error family, from Vertex AI project/location to WeChat keys — this error's family across 18 libraries.

Related errors

• `env.deployment` is a required configuration property for Az
• [Continue] Azure endpoint detected but API key appears to be
• Profile ${profileId} not found
• No reranker set up
• No chat model selected

AI-assisted analysis of continuedev/continue@5522c6f44c (2026-08-27). Data as JSON: /api/errors/0d74010db7413804. Report an issue: GitHub.