sync: upstream v0.83.0

.github/APPROVED_CONTRIBUTORS

@@ -295,3 +295,17 @@ christianbasch pr
 cpacker pr
 
 rgarcia pr
+
+renaudhartert-db pr
+
+HyeokjaeLee pr
+
+arajkumar pr
+
+brianstanley pr
+
+scruffymongrel pr
+
+SI-RUI-ZHANG pr
+
+sunnyyoung pr

.github/upstream.json

@@ -1,6 +1,6 @@
 {
 	"repo": "badlogic/pi-mono",
-	"tag": "v0.82.1",
-	"sha": "5bc1c2c0a6f07e00e8c240304182f213ab8d311f",
-	"synced_at": "2026-07-25T21:29:48Z"
+	"tag": "v0.83.0",
+	"sha": "71efc6f0c1909874ec8c944637a9ae7fc0e2d508",
+	"synced_at": "2026-07-30T03:38:57Z"
 }

packages/agent/src/proxy.ts

@@ -125,7 +125,7 @@ export function streamProxy(model: Model<any>, context: Context, options: ProxyS
 		// Initialize the partial message that we'll build up from events
 		const partial: AssistantMessage = {
 			role: "assistant",
-			stopReason: "stop",
+			stopReason: "pending",
 			content: [],
 			api: model.api,
 			provider: model.provider,

packages/ai/CHANGELOG.md

@@ -6,10 +6,25 @@
 
 ### Added
 
+- Added per-request `fetch` injection for supported text and image provider transports; Google adapters reject non-global implementations rather than silently bypassing them.
+- Added Claude Opus 5 support for the GitHub Copilot provider, routing through the Anthropic Messages API with adaptive thinking, 1M context, and the Copilot `minimal` thinking-level override ([#7158](https://github.com/earendil-works/pi/pull/7158) by [@jay-aye-see-kay](https://github.com/jay-aye-see-kay)).
+- Added the `"pending"` stop reason for partial streaming messages ([#7151](https://github.com/earendil-works/pi/pull/7151) by [@lucasmeijer](https://github.com/lucasmeijer)).
+- Added `AssistantMessage.rawStopReason` and populated it across Google, Anthropic, Amazon Bedrock, Mistral, and OpenAI streams; unmapped terminal reasons now surface as provider errors instead of successful stops ([#7272](https://github.com/earendil-works/pi/pull/7272)).
+- Added manual redirect URL and authorization-code entry to OpenRouter OAuth login for remote and headless environments ([#7114](https://github.com/earendil-works/pi/pull/7114) by [@rgarcia](https://github.com/rgarcia)).
+- Added `AuthResolutionOverrides.minOAuthValidityMs` so callers can require and refresh OAuth credentials with a minimum remaining validity ([#7168](https://github.com/earendil-works/pi/pull/7168)).
+
 ### Changed
 
+- Changed stored OAuth credentials to refresh when less than five minutes of validity remain instead of waiting until expiration ([#7168](https://github.com/earendil-works/pi/pull/7168)).
+
 ### Fixed
 
+- Fixed compiled validation of nullable array tool arguments with the inherited TypeBox validation update ([#7243](https://github.com/earendil-works/pi/pull/7243) by [@petrroll](https://github.com/petrroll)).
+- Fixed Qwen Token Plan reasoning models to send their service-specific thinking controls and supported reasoning-effort levels ([#6951](https://github.com/earendil-works/pi/issues/6951), [#6998](https://github.com/earendil-works/pi/issues/6998)).
+- Fixed Z.AI providers and compatible custom endpoints to send output limits through `max_tokens`, which those endpoints honor ([#7174](https://github.com/earendil-works/pi/pull/7174) by [@HyeokjaeLee](https://github.com/HyeokjaeLee)).
+- Fixed explicitly configured Amazon Bedrock profiles being overridden by ambient AWS access keys ([#7176](https://github.com/earendil-works/pi/pull/7176) by [@christianbasch](https://github.com/christianbasch)).
+- Fixed malformed OpenAI-compatible tool-call deltas with both a valid `function` payload and an empty `custom` object discarding the function arguments ([#7288](https://github.com/earendil-works/pi/pull/7288) by [@sunnyyoung](https://github.com/sunnyyoung)).
+
 ### Removed
 
 ## [2026.7.29-6] - 2026-07-29

packages/ai/README.md

@@ -874,7 +874,8 @@ for await (const event of s) {
 
 Every `AssistantMessage` includes a `stopReason` field that indicates how the generation ended:
 
-- `"stop"` - Normal completion, the model finished its response
+- `"pending"` - Only present in partial messages when we do not know what the stop reason will be
+- `"stop"` - This is the final message the model will produce this turn
 - `"length"` - Output hit the maximum token limit
 - `"toolUse"` - Model is calling tools and expects tool results
 - `"error"` - An error occurred during generation

packages/ai/scripts/generate-models.ts

@@ -265,6 +265,33 @@ const DEEPSEEK_V4_THINKING_LEVEL_MAP = {
 	high: "high",
 	max: "max",
 } as const;
+const QWEN_TOKEN_PLAN_HIGH_MAX_THINKING_LEVEL_MAP = {
+	minimal: null,
+	low: null,
+	medium: null,
+	high: "high",
+	xhigh: null,
+	max: "max",
+} as const;
+const QWEN_TOKEN_PLAN_QWEN38_THINKING_LEVEL_MAP = {
+	minimal: null,
+	low: "low",
+	medium: "medium",
+	high: null,
+	xhigh: "xhigh",
+	max: null,
+} as const;
+const QWEN_TOKEN_PLAN_REASONING_EFFORT_UNSUPPORTED_MODEL_IDS = new Set([
+	"MiniMax-M2.5",
+	"deepseek-v3.2",
+	"kimi-k2.5",
+	"kimi-k2.6",
+	"kimi-k2.7-code",
+	"qwen3.6-flash",
+	"qwen3.6-plus",
+	"qwen3.7-max",
+	"qwen3.7-plus",
+]);
 
 const KIMI_K3_THINKING_LEVEL_MAP = {
 	off: null,
@@ -430,6 +457,7 @@ const GITHUB_COPILOT_EXTENDED_CONTEXT_MODELS = new Set([
 	"claude-opus-4.6",
 	"claude-opus-4.7",
 	"claude-opus-4.8",
+	"claude-opus-5",
 	"claude-sonnet-4.6",
 	"claude-sonnet-5",
 	"gpt-5.3-codex",
@@ -442,6 +470,7 @@ const GITHUB_COPILOT_EXTENDED_CONTEXT_MODELS = new Set([
 const GITHUB_COPILOT_THINKING_LEVEL_OVERRIDES = {
 	"claude-opus-4.7": { minimal: "low" },
 	"claude-opus-4.8": { minimal: "low" },
+	"claude-opus-5": { minimal: "low" },
 	"claude-sonnet-4.6": { minimal: "low", max: "max" },
 } satisfies Record<string, NonNullable<Model<Api>["thinkingLevelMap"]>>;
 
@@ -630,7 +659,7 @@ function detectOpenAICompletionsCompat(model: Model<"openai-completions">): Open
 		isAntLing;
 
 	const useMaxTokens =
-		baseUrl.includes("chutes.ai") || isMoonshot || isCloudflareAiGateway || isTogether || isNvidia || isAntLing;
+		baseUrl.includes("chutes.ai") || isMoonshot || isCloudflareAiGateway || isTogether || isNvidia || isAntLing || isZai;
 
 	const isGrok = provider === "xai" || baseUrl.includes("api.x.ai");
 	const isDeepSeek = provider === "deepseek" || baseUrl.includes("deepseek.com");
@@ -2047,6 +2076,7 @@ async function loadModelsDevData(): Promise<Model<any>[]> {
 			thinkingFormat: "qwen",
 			supportsDeveloperRole: false,
 			supportsStore: false,
+			supportsReasoningEffort: true,
 		};
 		const qwenTokenPlanVariants = [
 			{
@@ -2068,14 +2098,25 @@ async function loadModelsDevData(): Promise<Model<any>[]> {
 			for (const [modelId, model] of Object.entries(providerModels)) {
 				const m = model as ModelsDevModel;
 				if (m.tool_call !== true) continue;
+				const supportsReasoningEffort = !QWEN_TOKEN_PLAN_REASONING_EFFORT_UNSUPPORTED_MODEL_IDS.has(modelId);
 
 				models.push({
 					id: modelId,
 					name: m.name || modelId,
 					api: "openai-completions",
 					provider,
 					baseUrl,
-					compat: qwenTokenPlanCompat,
+					compat: supportsReasoningEffort
+						? qwenTokenPlanCompat
+						: { ...qwenTokenPlanCompat, supportsReasoningEffort: false },
+					...(supportsReasoningEffort
+						? {
+								thinkingLevelMap:
+									modelId === "qwen3.8-max-preview"
+										? QWEN_TOKEN_PLAN_QWEN38_THINKING_LEVEL_MAP
+										: QWEN_TOKEN_PLAN_HIGH_MAX_THINKING_LEVEL_MAP,
+							}
+						: {}),
 					reasoning: m.reasoning === true,
 					input: m.modalities?.input?.includes("image") ? ["text", "image"] : ["text"],
 					cost: {
@@ -2576,7 +2617,12 @@ async function generateModels() {
 	allModels.push(...antLingModels);
 
 	for (const candidate of allModels) {
-		if (candidate.api === "openai-completions" && candidate.id.includes("deepseek-v4")) {
+		if (
+			candidate.api === "openai-completions" &&
+			candidate.id.includes("deepseek-v4") &&
+			candidate.provider !== "qwen-token-plan" &&
+			candidate.provider !== "qwen-token-plan-cn"
+		) {
 			const preservesNativeReasoningEffort = candidate.provider === "openrouter" || candidate.provider === "opencode";
 			candidate.compat = {
 				...candidate.compat,
@@ -2719,29 +2765,6 @@ async function generateModels() {
 		});
 	}
 
-	// Add qwen3.8-max-preview to Qwen Token Plan providers until models.dev includes it
-	for (const qwenTpProvider of ["qwen-token-plan", "qwen-token-plan-cn"] as const) {
-		if (!allModels.some((m) => m.provider === qwenTpProvider && m.id === "qwen3.8-max-preview")) {
-			const baseUrl =
-				qwenTpProvider === "qwen-token-plan"
-					? "https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1"
-					: "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1";
-			allModels.push({
-				id: "qwen3.8-max-preview",
-				name: "Qwen3.8 Max Preview",
-				api: "openai-completions",
-				provider: qwenTpProvider,
-				baseUrl,
-				compat: { thinkingFormat: "qwen", supportsDeveloperRole: false, supportsStore: false } satisfies OpenAICompletionsCompat,
-				reasoning: true,
-				input: ["text", "image"],
-				cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
-				contextWindow: 1000000,
-				maxTokens: 65536,
-			});
-		}
-	}
-
 	// Add "auto" alias for openrouter/auto
 	if (!allModels.some(m => m.provider === "openrouter" && m.id === "auto")) {
 		allModels.push({

packages/ai/src/api/anthropic-messages.ts

@@ -18,6 +18,7 @@ import type {
 	Message,
 	Model,
 	ProviderEnv,
+	ProviderHeaders,
 	ProviderNativeContent,
 	SimpleStreamOptions,
 	StopReason,
@@ -1199,7 +1200,7 @@ export const stream: StreamFunction<"anthropic-messages", AnthropicOptions> = (
 				totalTokens: 0,
 				cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
 			},
-			stopReason: "stop",
+			stopReason: "pending",
 			timestamp: Date.now(),
 		};
 
@@ -1243,6 +1244,7 @@ export const stream: StreamFunction<"anthropic-messages", AnthropicOptions> = (
 					options?.interleavedThinking ?? true,
 					shouldUseFineGrainedToolStreamingBeta(model, context),
 					optionsHeaders,
+					options?.fetch,
 					copilotDynamicHeaders,
 					cacheSessionId,
 					options?.env,
@@ -1489,6 +1491,7 @@ export const stream: StreamFunction<"anthropic-messages", AnthropicOptions> = (
 					}
 				} else if (event.type === "message_delta") {
 					if (event.delta.stop_reason) {
+						output.rawStopReason = event.delta.stop_reason;
 						const stopReasonResult = mapStopReason(event.delta.stop_reason, event.delta.stop_details);
 						output.stopReason = stopReasonResult.stopReason;
 						if (stopReasonResult.errorMessage) {
@@ -1537,6 +1540,9 @@ export const stream: StreamFunction<"anthropic-messages", AnthropicOptions> = (
 				applyServerFallbackAbort(output, serverFallbackReceipt);
 			}
 
+			if (output.stopReason === "pending") {
+				throw new Error("Anthropic stream ended without a stop reason");
+			}
 			if (output.stopReason === "aborted" || output.stopReason === "error") {
 				throw new Error(output.errorMessage || "An unknown error occurred");
 			}
@@ -1710,7 +1716,8 @@ function createClient(
 	apiKey: string | undefined,
 	interleavedThinking: boolean,
 	useFineGrainedToolStreamingBeta: boolean,
-	optionsHeaders?: Record<string, string>,
+	optionsHeaders?: ProviderHeaders,
+	fetch?: typeof globalThis.fetch,
 	dynamicHeaders?: Record<string, string>,
 	sessionId?: string,
 	env?: ProviderEnv,
@@ -1731,6 +1738,7 @@ function createClient(
 			authToken: null,
 			baseURL: resolveCloudflareBaseUrl(model, env),
 			dangerouslyAllowBrowser: true,
+			fetch,
 			defaultHeaders: sanitizeAdaptiveThinkingHeaders(
 				model,
 				mergeHeaders(
@@ -1758,6 +1766,7 @@ function createClient(
 			authToken: apiKey,
 			baseURL: model.baseUrl,
 			dangerouslyAllowBrowser: true,
+			fetch,
 			defaultHeaders: sanitizeAdaptiveThinkingHeaders(
 				model,
 				mergeHeaders(
@@ -1783,6 +1792,7 @@ function createClient(
 			authToken: apiKey,
 			baseURL: model.baseUrl,
 			dangerouslyAllowBrowser: true,
+			fetch,
 			defaultHeaders: sanitizeAdaptiveThinkingHeaders(
 				model,
 				mergeHeaders(
@@ -1810,6 +1820,7 @@ function createClient(
 		authToken: null,
 		baseURL: model.baseUrl,
 		dangerouslyAllowBrowser: true,
+		fetch,
 		defaultHeaders: sanitizeAdaptiveThinkingHeaders(
 			model,
 			mergeHeaders(
@@ -2350,7 +2361,11 @@ function mapStopReason(
 		case "stop_sequence":
 			return { stopReason: "stop" }; // We don't supply stop sequences, so this should never happen
 		case "sensitive": // Content flagged by safety filters (not yet in SDK types)
-			return { stopReason: "error", stopDetails: { type: "sensitive" } };
+			return {
+				stopReason: "error",
+				errorMessage: "Provider stopped with: sensitive",
+				stopDetails: { type: "sensitive" },
+			};
 		default:
 			// Handle unknown stop reasons gracefully (API may add new values)
 			throw new Error(`Unhandled stop reason: ${reason}`);

packages/ai/src/api/azure-openai-responses.ts

@@ -90,7 +90,7 @@ export const stream: StreamFunction<"azure-openai-responses", AzureOpenAIRespons
 				totalTokens: 0,
 				cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
 			},
-			stopReason: "stop",
+			stopReason: "pending",
 			timestamp: Date.now(),
 		};
 
@@ -132,6 +132,9 @@ export const stream: StreamFunction<"azure-openai-responses", AzureOpenAIRespons
 				throw new Error("Request was aborted");
 			}
 
+			if (output.stopReason === "pending") {
+				throw new Error("Azure OpenAI Responses stream ended without a stop reason");
+			}
 			if (output.stopReason === "aborted" || output.stopReason === "error") {
 				throw new Error("An unknown error occurred");
 			}
@@ -265,6 +268,7 @@ function createClient(model: Model<"azure-openai-responses">, apiKey: string, op
 		apiKey,
 		apiVersion,
 		dangerouslyAllowBrowser: true,
+		fetch: options?.fetch,
 		defaultHeaders: headers,
 		baseURL: baseUrl,
 	});

packages/ai/src/api/bedrock-converse-stream.ts

@@ -135,14 +135,20 @@ export const stream: StreamFunction<"bedrock-converse-stream", BedrockOptions> =
 				totalTokens: 0,
 				cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
 			},
-			stopReason: "stop",
+			stopReason: "pending",
 			timestamp: Date.now(),
 		};
 
 		const blocks = output.content as Block[];
 
+		// A profile explicitly configured through pi's auth flow (the `profile`
+		// option or scoped `AWS_PROFILE` on the stored credential's env) must win
+		// over ambient AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY. The SDK default
+		// chain already prefers a configured profile over env keys, but only when
+		// `credentials` is not set on the client config. See #6957.
+		const optionsProfile = options.profile || options.env?.AWS_PROFILE;
 		const config: BedrockRuntimeClientConfig = {
-			profile: options.profile || getProviderEnvValue("AWS_PROFILE", options.env),
+			profile: optionsProfile || getProviderEnvValue("AWS_PROFILE", options.env),
 		};
 		const configuredRegion = getConfiguredBedrockRegion(options);
 		const hasAmbientConfiguredProfile = Boolean(getProviderEnvValue("AWS_PROFILE"));
@@ -194,7 +200,7 @@ export const stream: StreamFunction<"bedrock-converse-stream", BedrockOptions> =
 			}
 
 			const credentials = getConfiguredBedrockCredentials(options.env);
-			if (!skipAuth && credentials) {
+			if (!skipAuth && credentials && !optionsProfile) {
 				config.credentials = credentials;
 			}
 
@@ -275,6 +281,7 @@ export const stream: StreamFunction<"bedrock-converse-stream", BedrockOptions> =
 				} else if (item.contentBlockStop) {
 					handleContentBlockStop(item.contentBlockStop, blocks, output, stream);
 				} else if (item.messageStop) {
+					output.rawStopReason = item.messageStop.stopReason;
 					const { stopReason, errorMessage } = mapStopReason(item.messageStop.stopReason);
 					output.stopReason = stopReason;
 					if (errorMessage) {
@@ -299,6 +306,9 @@ export const stream: StreamFunction<"bedrock-converse-stream", BedrockOptions> =
 				throw new Error("Request was aborted");
 			}
 
+			if (output.stopReason === "pending") {
+				throw new Error("Bedrock stream ended without a stop reason");
+			}
 			if (output.stopReason === "error" || output.stopReason === "aborted") {
 				throw new Error(output.errorMessage || "An unknown error occurred");
 			}
@@ -962,7 +972,9 @@ function mapStopReason(reason: string | undefined): { stopReason: StopReason; er
 		case BedrockStopReason.TOOL_USE:
 			return { stopReason: "toolUse" };
 		default:
-			return reason ? { stopReason: "error", errorMessage: reason } : { stopReason: "error" };
+			return reason
+				? { stopReason: "error", errorMessage: `Provider stopped with: ${reason}` }
+				: { stopReason: "error" };
 	}
 }