diff --git a/packages/ai/src/providers/openai-codex-responses.ts b/packages/ai/src/providers/openai-codex-responses.ts
index 2a15b7bc9..446727a9f 100644
--- a/packages/ai/src/providers/openai-codex-responses.ts
+++ b/packages/ai/src/providers/openai-codex-responses.ts
@@ -104,7 +104,7 @@ import { transformMessages } from "./transform-messages";
 export interface OpenAICodexResponsesOptions extends StreamOptions {
 	reasoning?: "none" | "minimal" | "low" | "medium" | "high" | "xhigh";
 	reasoningSummary?: "auto" | "concise" | "detailed" | null;
-	/** `reasoning.context` replay scope; defaults to `all_turns` for every Codex request when unset. */
+	/** `reasoning.context` replay scope; defaults to `all_turns` when unset. The `all_turns` value is gated to gpt-5.4+ Codex models — older ids reject it, so it is suppressed and `context` omitted. */
 	reasoningContext?: CodexReasoningContext;
 	textVerbosity?: "low" | "medium" | "high";
 	include?: string[];
diff --git a/packages/ai/src/providers/openai-codex/request-transformer.ts b/packages/ai/src/providers/openai-codex/request-transformer.ts
index b627fcbfa..77882a60e 100644
--- a/packages/ai/src/providers/openai-codex/request-transformer.ts
+++ b/packages/ai/src/providers/openai-codex/request-transformer.ts
@@ -1,4 +1,5 @@
 import type { Effort } from "@oh-my-pi/pi-catalog/effort";
+import { supportsAllTurnsReasoningContext } from "@oh-my-pi/pi-catalog/identity";
 import { requireSupportedEffort } from "@oh-my-pi/pi-catalog/model-thinking";
 import type { Api, Model } from "../../types";
 
@@ -14,7 +15,7 @@ export interface ReasoningConfig {
 export interface CodexRequestOptions {
 	reasoningEffort?: ReasoningConfig["effort"];
 	reasoningSummary?: ReasoningConfig["summary"] | null;
-	/** Explicit `reasoning.context` override; defaults to `all_turns` for every Codex request when unset. */
+	/** Explicit `reasoning.context` override; defaults to `all_turns` when unset. The `all_turns` value is gated to gpt-5.4+ Codex models — older ids reject it, so it is suppressed and `context` omitted. */
 	reasoningContext?: CodexReasoningContext;
 	textVerbosity?: "low" | "medium" | "high";
 	include?: string[];
@@ -254,9 +255,21 @@ export async function transformRequestBody(
 			...body.reasoning,
 			...reasoningConfig,
 		};
-		// Default reasoning replay to `all_turns` for every Codex request,
-		// mirroring codex-rs; an explicit `reasoningContext` overrides it.
-		body.reasoning.context = options.reasoningContext ?? "all_turns";
+		// Default reasoning replay to `all_turns`, mirroring codex-rs; an
+		// explicit `reasoningContext` overrides the default. The `all_turns`
+		// value is only accepted from gpt-5.4 onward — earlier Codex ids
+		// (gpt-5.1-codex, gpt-5.3-codex, gpt-5.3-codex-spark) reject it with
+		// "Unsupported value: 'all_turns' is not supported with this model".
+		// For those, drop `context` so the server applies its `current_turn`
+		// default. The version gate is authoritative: even an explicit
+		// `all_turns` override is suppressed on unsupported models, while
+		// `current_turn`/`auto` (universally supported) always pass through.
+		const context = options.reasoningContext ?? "all_turns";
+		if (context === "all_turns" && !supportsAllTurnsReasoningContext(model.id)) {
+			delete body.reasoning.context;
+		} else {
+			body.reasoning.context = context;
+		}
 	} else {
 		delete body.reasoning;
 	}
diff --git a/packages/ai/test/openai-codex-responses-lite.test.ts b/packages/ai/test/openai-codex-responses-lite.test.ts
index 575c3cdd3..c6aee594f 100644
--- a/packages/ai/test/openai-codex-responses-lite.test.ts
+++ b/packages/ai/test/openai-codex-responses-lite.test.ts
@@ -83,21 +83,21 @@ function createCodexFetchMock(sse: string, onRequest: (captured: CapturedCodexRe
 }
 
 describe("openai-codex reasoning.context", () => {
-	it("forwards an explicit reasoning.context and defaults to all_turns", async () => {
-		const model = createCodexModel("gpt-5.1-codex");
+	it("defaults to all_turns on gpt-5.4+ models and forwards explicit overrides", async () => {
+		const model = createCodexModel("gpt-5.4");
+
+		const defaulted = await transformRequestBody({ model: model.id }, model, { reasoningEffort: "medium" });
+		expect(defaulted.reasoning?.context).toBe("all_turns");
 
 		const explicit = await transformRequestBody({ model: model.id }, model, {
 			reasoningEffort: "medium",
 			reasoningContext: "current_turn",
 		});
 		expect(explicit.reasoning?.context).toBe("current_turn");
-
-		const defaulted = await transformRequestBody({ model: model.id }, model, { reasoningEffort: "medium" });
-		expect(defaulted.reasoning?.context).toBe("all_turns");
 	});
 
-	it("defaults reasoning.context to all_turns under Responses Lite unless overridden", async () => {
-		const model = createCodexModel("gpt-5.1-codex");
+	it("keeps the all_turns default for the lite transport on supported models", async () => {
+		const model = createCodexModel("gpt-5.5");
 
 		const lite = await transformRequestBody({ model: model.id }, model, {
 			reasoningEffort: "medium",
@@ -112,6 +112,39 @@ describe("openai-codex reasoning.context", () => {
 		});
 		expect(overridden.reasoning?.context).toBe("auto");
 	});
+
+	// gpt-5.1-codex / gpt-5.3-codex / gpt-5.3-codex-spark reject `all_turns`
+	// ("Unsupported value: 'all_turns' is not supported with this model").
+	it.each([
+		"gpt-5.1-codex",
+		"gpt-5.3-codex",
+		"gpt-5.3-codex-spark",
+	])("omits the all_turns default for pre-5.4 model %s", async modelId => {
+		const model = createCodexModel(modelId);
+
+		const defaulted = await transformRequestBody({ model: model.id }, model, { reasoningEffort: "medium" });
+		expect(defaulted.reasoning).toBeDefined();
+		expect(defaulted.reasoning?.context).toBeUndefined();
+		expect("context" in (defaulted.reasoning ?? {})).toBe(false);
+
+		// A supported override (current_turn/auto) is still honored.
+		const overridden = await transformRequestBody({ model: model.id }, model, {
+			reasoningEffort: "medium",
+			reasoningContext: "current_turn",
+		});
+		expect(overridden.reasoning?.context).toBe("current_turn");
+	});
+
+	it("suppresses an explicit all_turns override on a pre-5.4 model", async () => {
+		const model = createCodexModel("gpt-5.3-codex-spark");
+
+		const forced = await transformRequestBody({ model: model.id }, model, {
+			reasoningEffort: "medium",
+			reasoningContext: "all_turns",
+		});
+		expect(forced.reasoning).toBeDefined();
+		expect(forced.reasoning?.context).toBeUndefined();
+	});
 });
 
 describe("openai-codex Responses Lite input shaping", () => {
diff --git a/packages/catalog/src/identity/family.ts b/packages/catalog/src/identity/family.ts
index c48d91f35..031a77c31 100644
--- a/packages/catalog/src/identity/family.ts
+++ b/packages/catalog/src/identity/family.ts
@@ -13,6 +13,7 @@ import {
 	parseAnthropicModel,
 	parseGlmModel,
 	parseKnownModel,
+	parseOpenAIModel,
 	semverGte,
 } from "./classify";
 
@@ -121,6 +122,22 @@ export const isOpenAIModelId = memo((modelId: string): boolean => {
 	return /(^|\/)(gpt|o1|o3|o4)[-.]/i.test(modelId) || modelId.toLowerCase().includes("openai/");
 });
 
+/**
+ * OpenAI Codex models that honor `reasoning.context: "all_turns"` (full
+ * cross-turn reasoning replay). The `reasoning.context` field itself exists for
+ * the whole gpt-5/o-series family, but the `all_turns` value is only accepted
+ * from gpt-5.4 onward; earlier ids (`gpt-5.1-codex`, `gpt-5.3-codex`, and
+ * `gpt-5.3-codex-spark`) reject it with
+ * `Unsupported value: 'all_turns' is not supported with this model`. Version
+ * floor (not an allowlist) so 5.6/6.x inherit support automatically. Callers
+ * fall back to omitting `context`, letting the server default to `current_turn`.
+ */
+export const supportsAllTurnsReasoningContext = memo((modelId: string): boolean => {
+	const parsed = parseOpenAIModel(bareModelId(modelId));
+	if (!parsed) return false;
+	return semverGte(parsed.version, "5.4");
+});
+
 /**
  * Reasoning-capable GLM coding SKUs: glm-4.5 and up on the base / `-air` /
  * `-turbo` lines. Excludes the vision (`…v`) shape, the non-reasoning