From f38592f1bb029b163bde06bcecae8cab4fd0d3e3 Mon Sep 17 00:00:00 2001
From: Pugazhendhi <pugazhendhi@unboundsecurity.ai>
Date: Thu, 12 Jun 2025 10:37:37 +0530
Subject: [PATCH 1/2] Add cache breakpoints for custom vertex models on Unbound

---
 src/api/providers/fetchers/unbound.ts | 2 +-
 src/api/providers/unbound.ts          | 5 +++++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/src/api/providers/fetchers/unbound.ts b/src/api/providers/fetchers/unbound.ts
index 98c0c58fa5..202a1b6cb1 100644
--- a/src/api/providers/fetchers/unbound.ts
+++ b/src/api/providers/fetchers/unbound.ts
@@ -20,7 +20,7 @@ export async function getUnboundModels(apiKey?: string | null): Promise<Record<s
 			for (const [modelId, model] of Object.entries(rawModels)) {
 				const modelInfo: ModelInfo = {
 					maxTokens: model?.maxTokens ? parseInt(model.maxTokens) : undefined,
-					contextWindow: model?.contextWindow ? parseInt(model.contextWindow) : 0,
+					contextWindow: model?.contextWindow ? parseInt(model.contextWindow) : 128_000,
 					supportsImages: model?.supportsImages ?? false,
 					supportsPromptCache: model?.supportsPromptCaching ?? false,
 					supportsComputerUse: model?.supportsComputerUse ?? false,
diff --git a/src/api/providers/unbound.ts b/src/api/providers/unbound.ts
index 2c7bd1e575..bc85dfd499 100644
--- a/src/api/providers/unbound.ts
+++ b/src/api/providers/unbound.ts
@@ -9,6 +9,7 @@ import { ApiStream, ApiStreamUsageChunk } from "../transform/stream"
 import { convertToOpenAiMessages } from "../transform/openai-format"
 import { addCacheBreakpoints as addAnthropicCacheBreakpoints } from "../transform/caching/anthropic"
 import { addCacheBreakpoints as addGeminiCacheBreakpoints } from "../transform/caching/gemini"
+import { addCacheBreakpoints as addVertexCacheBreakpoints } from "../transform/caching/vertex"
 
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
 import { RouterProvider } from "./router-provider"
@@ -70,6 +71,10 @@ export class UnboundHandler extends RouterProvider implements SingleCompletionHa
 				addAnthropicCacheBreakpoints(systemPrompt, openAiMessages)
 			}
 		}
+		// Custom models from Vertex AI (no configuration) need to be handled differently.
+		if (modelId.startsWith("vertex-ai/google.") || modelId.startsWith("vertex-ai/anthropic.")) {
+			addVertexCacheBreakpoints(messages)
+		}
 
 		// Required by Anthropic; other providers default to max tokens allowed.
 		let maxTokens: number | undefined

From 6871ae53b41d6d7e054281f978b7859a32b62bee Mon Sep 17 00:00:00 2001
From: Pugazhendhi <pugazhendhi@unboundsecurity.ai>
Date: Thu, 12 Jun 2025 12:16:08 +0530
Subject: [PATCH 2/2] Reverts context window

---
 src/api/providers/fetchers/unbound.ts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/api/providers/fetchers/unbound.ts b/src/api/providers/fetchers/unbound.ts
index 202a1b6cb1..98c0c58fa5 100644
--- a/src/api/providers/fetchers/unbound.ts
+++ b/src/api/providers/fetchers/unbound.ts
@@ -20,7 +20,7 @@ export async function getUnboundModels(apiKey?: string | null): Promise<Record<s
 			for (const [modelId, model] of Object.entries(rawModels)) {
 				const modelInfo: ModelInfo = {
 					maxTokens: model?.maxTokens ? parseInt(model.maxTokens) : undefined,
-					contextWindow: model?.contextWindow ? parseInt(model.contextWindow) : 128_000,
+					contextWindow: model?.contextWindow ? parseInt(model.contextWindow) : 0,
 					supportsImages: model?.supportsImages ?? false,
 					supportsPromptCache: model?.supportsPromptCaching ?? false,
 					supportsComputerUse: model?.supportsComputerUse ?? false,