feat(bedrock): support Claude Opus 4.8 (extends 4.7 adaptive thinking detection)

- Anthropic provider: add claude-opus-4-8 to both prompt-caching switch
  statements so it gets the same handling as 4.7 (native 1M context, no beta
  header required).
- Bedrock provider: rename isGen47Model -> isAdaptiveThinkingModel and expand
  the pattern to match opus-4-7, opus-4-8, sonnet-4-7, sonnet-4-8. 4.8 inherits
  the same adaptive-thinking + temperature-rejection contract from 4.7 with no
  breaking API changes.
- OpenAI-compatible provider: update comment to mention 4.8 alongside 4.7; no
  logic change (already honors the supportsTemperature: false flag).

The rename describes the capability (adaptive thinking) rather than a specific
generation, making future Claude versions easier to support.

Author: vandre-sales

src/api/providers/anthropic.ts

@@ -93,6 +93,7 @@ export class AnthropicHandler extends BaseProvider implements SingleCompletionHa
 			case "claude-sonnet-4-20250514":
 			case "claude-opus-4-6":
 			case "claude-opus-4-7":
+			case "claude-opus-4-8":
 			case "claude-opus-4-5-20251101":
 			case "claude-opus-4-1-20250805":
 			case "claude-opus-4-20250514":
@@ -161,6 +162,7 @@ export class AnthropicHandler extends BaseProvider implements SingleCompletionHa
 								case "claude-sonnet-4-20250514":
 								case "claude-opus-4-6":
 								case "claude-opus-4-7":
+								case "claude-opus-4-8":
 								case "claude-opus-4-5-20251101":
 								case "claude-opus-4-1-20250805":
 								case "claude-opus-4-20250514":

src/api/providers/bedrock.ts

@@ -392,10 +392,19 @@ export class AwsBedrockHandler extends BaseProvider implements SingleCompletionH
 		let additionalModelRequestFields: BedrockAdditionalModelFields | undefined
 		let thinkingEnabled = false
 
-		// Detect model generation for API compatibility
-		// Claude 4.7+ removed sampling params (temperature/top_p/top_k) and uses adaptive thinking
+		// Detect models that require the adaptive-thinking API contract.
+		// Starting with Claude Opus 4.7 (and the matching Sonnet 4.7), and continuing
+		// in Opus 4.8 / Sonnet 4.8, Anthropic removed sampling parameters
+		// (temperature/top_p/top_k) and replaced budget_tokens-based thinking with
+		// `thinking.type: "adaptive"` plus `output_config.effort`. The migration guide
+		// from 4.7 → 4.8 confirms there are no further breaking API changes, so we
+		// keep a single guard here that matches both generations.
 		const baseModelId = this.parseBaseModelId(modelConfig.id)
-		const isGen47Model = baseModelId.includes("opus-4-7") || baseModelId.includes("sonnet-4-7")
+		const isAdaptiveThinkingModel =
+			baseModelId.includes("opus-4-7") ||
+			baseModelId.includes("opus-4-8") ||
+			baseModelId.includes("sonnet-4-7") ||
+			baseModelId.includes("sonnet-4-8")
 
 		// Determine if thinking should be enabled
 		// metadata?.thinking?.enabled: Explicitly enabled through API metadata (direct request)
@@ -408,9 +417,13 @@ export class AwsBedrockHandler extends BaseProvider implements SingleCompletionH
 
 		if ((isThinkingExplicitlyEnabled || isThinkingEnabledBySettings) && modelConfig.info.supportsReasoningBudget) {
 			thinkingEnabled = true
-			if (isGen47Model) {
-				// Claude 4.7+ uses adaptive thinking with effort levels — budget_tokens causes 400 error
-				// display: "summarized" surfaces thinking content in Zoo Code UI
+			if (isAdaptiveThinkingModel) {
+				// Claude 4.7+ (incl. 4.8) uses adaptive thinking with effort levels —
+				// budget_tokens causes a 400 error.
+				// display: "summarized" surfaces thinking content in Zoo Code UI.
+				// effort "xhigh" remains the recommended level for agentic coding tasks
+				// across both 4.7 and 4.8 (4.8 changed the API default to "high" but
+				// the model continues to honour "xhigh" for deeper reasoning).
 				additionalModelRequestFields = {
 					thinking: { type: "adaptive", display: "summarized" },
 					output_config: { effort: "xhigh" },
@@ -432,8 +445,9 @@ export class AwsBedrockHandler extends BaseProvider implements SingleCompletionH
 
 		const inferenceConfig: BedrockInferenceConfig = {
 			maxTokens: modelConfig.maxTokens || (modelConfig.info.maxTokens as number),
-			// Claude 4.7+ removed temperature parameter entirely — causes 400 error if sent
-			...(isGen47Model
+			// Claude 4.7+ (including 4.8) removed sampling parameters entirely —
+			// sending temperature causes a 400 error.
+			...(isAdaptiveThinkingModel
 				? {}
 				: { temperature: modelConfig.temperature ?? (this.options.modelTemperature as number) }),
 		}

src/api/providers/openai.ts

@@ -154,9 +154,10 @@ export class OpenAiHandler extends BaseProvider implements SingleCompletionHandl
 
 			const requestOptions: OpenAI.Chat.Completions.ChatCompletionCreateParamsStreaming = {
 				model: modelId,
-				// Some OpenAI-Compatible models (e.g. claude-opus-4-7) reject `temperature` as
-				// deprecated/unsupported. Honor the model's `supportsTemperature` flag and omit it
-				// when explicitly set to false (undefined still sends temperature, preserving behavior).
+				// Some OpenAI-Compatible models (e.g. claude-opus-4-7, claude-opus-4-8) reject
+				// `temperature` as deprecated/unsupported. Honor the model's `supportsTemperature`
+				// flag and omit it when explicitly set to false (undefined still sends temperature,
+				// preserving behavior).
 				...(modelInfo.supportsTemperature !== false && {
 					temperature:
 						this.options.modelTemperature ?? (deepseekReasoner ? DEEP_SEEK_DEFAULT_TEMPERATURE : 0),