Add supportsUsageInStreaming compat flag for OpenAI-compatible providers

Renamed from supportsStreamOptions to clarify this controls stream_options: { include_usage: true }.
Defaults to true (no behavioral change for existing providers).
Providers like gatewayz.ai that reject this parameter can set supportsUsageInStreaming: false in model config.

Based on #596 by @XesGaDeus

packages/ai/CHANGELOG.md

@@ -2,8 +2,8 @@
 
 ## [Unreleased]
 
-### Fixed
-- Fixed 422 error with non-standard providers (gatewayz.ai, chutes.ai, etc.) by adding `supportsStreamOptions` compatibility flag and conditionally sending `stream_options` parameter.
+### Added
+- Added `supportsUsageInStreaming` compatibility flag for OpenAI-compatible providers that reject `stream_options: { include_usage: true }`. Defaults to `true`. Set to `false` in model config for providers like gatewayz.ai. ([#596](https://github.com/badlogic/pi-mono/pull/596) by [@XesGaDeus](https://github.com/XesGaDeus))
 
 ## [0.42.0] - 2026-01-09
 

packages/ai/src/providers/openai-completions.ts

@@ -369,7 +369,7 @@ function buildParams(model: Model<"openai-completions">, context: Context, optio
 		stream: true,
 	};
 
-	if (compat.supportsStreamOptions) {
+	if (compat.supportsUsageInStreaming !== false) {
 		(params as any).stream_options = { include_usage: true };
 	}
 
@@ -644,8 +644,7 @@ function detectCompatFromUrl(baseUrl: string): Required<OpenAICompat> {
 		baseUrl.includes("cerebras.ai") ||
 		baseUrl.includes("api.x.ai") ||
 		baseUrl.includes("mistral.ai") ||
-		baseUrl.includes("chutes.ai") ||
-		baseUrl.includes("gatewayz.ai");
+		baseUrl.includes("chutes.ai");
 
 	const useMaxTokens = baseUrl.includes("mistral.ai") || baseUrl.includes("chutes.ai");
 
@@ -657,7 +656,7 @@ function detectCompatFromUrl(baseUrl: string): Required<OpenAICompat> {
 		supportsStore: !isNonStandard,
 		supportsDeveloperRole: !isNonStandard,
 		supportsReasoningEffort: !isGrok,
-		supportsStreamOptions: !isNonStandard,
+		supportsUsageInStreaming: true,
 		maxTokensField: useMaxTokens ? "max_tokens" : "max_completion_tokens",
 		requiresToolResultName: isMistral,
 		requiresAssistantAfterToolResult: false, // Mistral no longer requires this as of Dec 2024
@@ -678,7 +677,7 @@ function getCompat(model: Model<"openai-completions">): Required<OpenAICompat> {
 		supportsStore: model.compat.supportsStore ?? detected.supportsStore,
 		supportsDeveloperRole: model.compat.supportsDeveloperRole ?? detected.supportsDeveloperRole,
 		supportsReasoningEffort: model.compat.supportsReasoningEffort ?? detected.supportsReasoningEffort,
-		supportsStreamOptions: model.compat.supportsStreamOptions ?? detected.supportsStreamOptions,
+		supportsUsageInStreaming: model.compat.supportsUsageInStreaming ?? detected.supportsUsageInStreaming,
 		maxTokensField: model.compat.maxTokensField ?? detected.maxTokensField,
 		requiresToolResultName: model.compat.requiresToolResultName ?? detected.requiresToolResultName,
 		requiresAssistantAfterToolResult:

packages/ai/src/types.ts

@@ -207,8 +207,8 @@ export interface OpenAICompat {
 	supportsDeveloperRole?: boolean;
 	/** Whether the provider supports `reasoning_effort`. Default: auto-detected from URL. */
 	supportsReasoningEffort?: boolean;
-	/** Whether the provider supports `stream_options`. Default: auto-detected from URL. */
-	supportsStreamOptions?: boolean;
+	/** Whether the provider supports `stream_options: { include_usage: true }` for token usage in streaming responses. Default: true. */
+	supportsUsageInStreaming?: boolean;
 	/** Which field to use for max tokens. Default: auto-detected from URL. */
 	maxTokensField?: "max_completion_tokens" | "max_tokens";
 	/** Whether tool results require the `name` field. Default: auto-detected from URL. */