The open source coding agent.
TypeScript
feat: enable prompt caching and cache token tracking for google-vertex-anthropic (#20266)
Major Hayden26cc924ea223
diff --git a/packages/opencode/src/provider/transform.ts b/packages/opencode/src/provider/transform.tsindex 219200b7658..c402238685f 100644--- a/packages/opencode/src/provider/transform.ts+++ b/packages/opencode/src/provider/transform.ts@@ -280,6 +280,7 @@ export namespace ProviderTransform {msgs = normalizeMessages(msgs, model, options)if ((model.providerID === "anthropic" ||+ model.providerID === "google-vertex-anthropic" ||model.api.id.includes("anthropic") ||model.api.id.includes("claude") ||model.id.includes("anthropic") ||diff --git a/packages/opencode/src/session/index.ts b/packages/opencode/src/session/index.tsindex 74506c31da6..94aee14c09c 100644--- a/packages/opencode/src/session/index.ts+++ b/packages/opencode/src/session/index.ts@@ -257,6 +257,9 @@ export namespace Session {const cacheReadInputTokens = safe(input.usage.cachedInputTokens ?? 0)const cacheWriteInputTokens = safe((input.metadata?.["anthropic"]?.["cacheCreationInputTokens"] ??+ // google-vertex-anthropic returns metadata under "vertex" key+ // (AnthropicMessagesLanguageModel custom provider key from 'vertex.anthropic.messages')+ input.metadata?.["vertex"]?.["cacheCreationInputTokens"] ??// @ts-expect-errorinput.metadata?.["bedrock"]?.["usage"]?.["cacheWriteInputTokens"] ??// @ts-expect-errordiff --git a/packages/opencode/test/provider/transform.test.ts b/packages/opencode/test/provider/transform.test.tsindex 5a1016bb7d7..0aee396f44a 100644--- a/packages/opencode/test/provider/transform.test.ts+++ b/packages/opencode/test/provider/transform.test.ts@@ -1792,6 +1792,58 @@ describe("ProviderTransform.message - cache control on gateway", () => {},})})++ test("google-vertex-anthropic applies cache control", () => {+ const model = createModel({+ providerID: "google-vertex-anthropic",+ api: {+ id: "google-vertex-anthropic",+ url: "https://us-central1-aiplatform.googleapis.com",+ npm: "@ai-sdk/google-vertex/anthropic",+ },+ id: "claude-sonnet-4@20250514",+ })+ const msgs = [+ {+ role: "system",+ content: "You are a helpful assistant",+ },+ {+ role: "user",+ content: "Hello",+ },+ ] as any[]++ const result = ProviderTransform.message(msgs, model, {}) as any[]++ expect(result[0].providerOptions).toEqual({+ anthropic: {+ cacheControl: {+ type: "ephemeral",+ },+ },+ openrouter: {+ cacheControl: {+ type: "ephemeral",+ },+ },+ bedrock: {+ cachePoint: {+ type: "default",+ },+ },+ openaiCompatible: {+ cache_control: {+ type: "ephemeral",+ },+ },+ copilot: {+ copilot_cache_control: {+ type: "ephemeral",+ },+ },+ })+ })})describe("ProviderTransform.variants", () => {diff --git a/packages/opencode/test/session/compaction.test.ts b/packages/opencode/test/session/compaction.test.tsindex a686d7ccffa..e6d715728d3 100644--- a/packages/opencode/test/session/compaction.test.ts+++ b/packages/opencode/test/session/compaction.test.ts@@ -1199,4 +1199,26 @@ describe("session.getUsage", () => {expect(result.tokens.total).toBe(1500)},)++ test("extracts cache write tokens from vertex metadata key", () => {+ const model = createModel({ context: 100_000, output: 32_000, npm: "@ai-sdk/google-vertex/anthropic" })+ const result = Session.getUsage({+ model,+ usage: {+ inputTokens: 1000,+ outputTokens: 500,+ totalTokens: 1500,+ cachedInputTokens: 200,+ },+ metadata: {+ vertex: {+ cacheCreationInputTokens: 300,+ },+ },+ })++ expect(result.tokens.input).toBe(500)+ expect(result.tokens.cache.read).toBe(200)+ expect(result.tokens.cache.write).toBe(300)+ })})