diff --git a/packages/llm/src/auth.ts b/packages/llm/src/auth.ts
index 5b3474abfe..3235f9d8dd 100644
--- a/packages/llm/src/auth.ts
+++ b/packages/llm/src/auth.ts
@@ -10,7 +10,8 @@ import type { LLMError, LLMRequest } from "./schema"
  * Most adapters use the default `Auth.bearer`, which reads
  * `request.model.apiKey` and sets `Authorization: Bearer ...`. Providers
  * that use a different header pick `Auth.apiKeyHeader(name)` (e.g.
- * Anthropic's `x-api-key`, Gemini's `x-goog-api-key`).
+ * Anthropic's `x-api-key`, Gemini's `x-goog-api-key`) or a provider-aware
+ * helper such as `Auth.openAI` for Azure OpenAI's static `api-key` header.
  *
  * Adapters that need per-request signing (AWS SigV4, future Vertex IAM,
  * future Azure AAD) implement `Auth` as a function that hashes the body,
@@ -52,6 +53,22 @@ const fromApiKey = (from: (apiKey: string) => Record<string, string>): Auth => (
  */
 export const bearer: Auth = fromApiKey((key) => ({ authorization: `Bearer ${key}` }))
 
+/**
+ * OpenAI-compatible auth with Azure OpenAI's static API-key exception. Azure
+ * Entra/OAuth callers can still pre-set `authorization` and omit `apiKey`.
+ */
+export const openAI: Auth = ({ request, headers }) => {
+  const key = request.model.apiKey
+  if (!key) return Effect.succeed(headers)
+  if (request.model.provider === "azure") {
+    return Effect.succeed({
+      ...Object.fromEntries(Object.entries(headers).filter(([name]) => name.toLowerCase() !== "authorization")),
+      "api-key": key,
+    })
+  }
+  return Effect.succeed({ ...headers, authorization: `Bearer ${key}` })
+}
+
 /**
  * Set a custom header to `request.model.apiKey`. No-op when `model.apiKey`
  * is unset. Used by Anthropic (`x-api-key`) and Gemini (`x-goog-api-key`).
diff --git a/packages/llm/src/provider/openai-chat.ts b/packages/llm/src/provider/openai-chat.ts
index aca8c284f2..772bcdb6e4 100644
--- a/packages/llm/src/provider/openai-chat.ts
+++ b/packages/llm/src/provider/openai-chat.ts
@@ -1,5 +1,6 @@
 import { Effect, Schema } from "effect"
 import { Adapter } from "../adapter"
+import { Auth } from "../auth"
 import { Endpoint } from "../endpoint"
 import { Framing } from "../framing"
 import { capabilities, model as llmModel, type ModelInput } from "../llm"
@@ -355,6 +356,7 @@ export const adapter = Adapter.fromProtocol({
   id: ADAPTER,
   protocol,
   endpoint: Endpoint.baseURL({ default: "https://api.openai.com/v1", path: "/chat/completions" }),
+  auth: Auth.openAI,
   framing: Framing.sse,
 })
 
diff --git a/packages/llm/src/provider/openai-responses.ts b/packages/llm/src/provider/openai-responses.ts
index 1dd8474b4d..37d29a27bf 100644
--- a/packages/llm/src/provider/openai-responses.ts
+++ b/packages/llm/src/provider/openai-responses.ts
@@ -1,5 +1,6 @@
 import { Effect, Schema } from "effect"
 import { Adapter } from "../adapter"
+import { Auth } from "../auth"
 import { Endpoint } from "../endpoint"
 import { Framing } from "../framing"
 import { capabilities, model as llmModel, type ModelInput } from "../llm"
@@ -385,6 +386,7 @@ export const adapter = Adapter.fromProtocol({
   id: ADAPTER,
   protocol,
   endpoint: Endpoint.baseURL({ default: "https://api.openai.com/v1", path: "/responses" }),
+  auth: Auth.openAI,
   framing: Framing.sse,
 })
 
diff --git a/packages/llm/test/provider/openai-chat.test.ts b/packages/llm/test/provider/openai-chat.test.ts
index 89e0c3486d..5a05508a74 100644
--- a/packages/llm/test/provider/openai-chat.test.ts
+++ b/packages/llm/test/provider/openai-chat.test.ts
@@ -83,6 +83,37 @@ describe("OpenAI Chat adapter", () => {
     }),
   )
 
+  it.effect("uses Azure api-key header for static OpenAI Chat keys", () =>
+    Effect.gen(function* () {
+      yield* LLMClient.make({ adapters: [OpenAIChat.adapter] })
+        .generate(
+          LLM.updateRequest(request, {
+            model: LLM.model({
+              ...model,
+              provider: "azure",
+              baseURL: "https://opencode-test.openai.azure.com/openai/v1/",
+              apiKey: "azure-key",
+              headers: { authorization: "Bearer stale" },
+            }),
+          }),
+        )
+        .pipe(
+          Effect.provide(
+            dynamicResponse((input) =>
+              Effect.gen(function* () {
+                const web = yield* HttpClientRequest.toWeb(input.request).pipe(Effect.orDie)
+                expect(web.headers.get("api-key")).toBe("azure-key")
+                expect(web.headers.get("authorization")).toBeNull()
+                return input.respond(sseEvents(deltaChunk({}, "stop")), {
+                  headers: { "content-type": "text/event-stream" },
+                })
+              }),
+            ),
+          ),
+        )
+    }),
+  )
+
   it.effect("prepares assistant tool-call and tool-result messages", () =>
     Effect.gen(function* () {
       const prepared = yield* LLMClient.make({ adapters: [OpenAIChat.adapter] }).prepare(
diff --git a/packages/llm/test/provider/openai-responses.test.ts b/packages/llm/test/provider/openai-responses.test.ts
index 255f622d62..b697d267c5 100644
--- a/packages/llm/test/provider/openai-responses.test.ts
+++ b/packages/llm/test/provider/openai-responses.test.ts
@@ -62,6 +62,37 @@ describe("OpenAI Responses adapter", () => {
     }),
   )
 
+  it.effect("uses Azure api-key header for static OpenAI Responses keys", () =>
+    Effect.gen(function* () {
+      yield* LLMClient.make({ adapters: [OpenAIResponses.adapter] })
+        .generate(
+          LLM.updateRequest(request, {
+            model: LLM.model({
+              ...model,
+              provider: "azure",
+              baseURL: "https://opencode-test.openai.azure.com/openai/v1/",
+              apiKey: "azure-key",
+              headers: { authorization: "Bearer stale" },
+            }),
+          }),
+        )
+        .pipe(
+          Effect.provide(
+            dynamicResponse((input) =>
+              Effect.gen(function* () {
+                const web = yield* HttpClientRequest.toWeb(input.request).pipe(Effect.orDie)
+                expect(web.headers.get("api-key")).toBe("azure-key")
+                expect(web.headers.get("authorization")).toBeNull()
+                return input.respond(sseEvents({ type: "response.completed", response: {} }), {
+                  headers: { "content-type": "text/event-stream" },
+                })
+              }),
+            ),
+          ),
+        )
+    }),
+  )
+
   it.effect("prepares function call and function output input items", () =>
     Effect.gen(function* () {
       const prepared = yield* LLMClient.make({ adapters: [OpenAIResponses.adapter] }).prepare(