From 91cd169eb3dbe15c568706e9fee470c7db8e3303 Mon Sep 17 00:00:00 2001
From: Ammar <ammar+ai@ammar.io>
Date: Thu, 12 Feb 2026 18:58:04 -0600
Subject: [PATCH] =?UTF-8?q?=F0=9F=A4=96=20fix:=20add=20GPT-5.3-Codex-Spark?=
 =?UTF-8?q?=20model=20data=20override?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Add a models-extra override for gpt-5.3-codex-spark using the newly published
Codex metadata (128k context and text-only capabilities), and add focused tests
for model stats and capability lookup behavior.

---

_Generated with `mux` • Model: `openai:gpt-5.3-codex` • Thinking: `xhigh` • Cost: `$0.00`_

<!-- mux-attribution: model=openai:gpt-5.3-codex thinking=xhigh costs=0.00 -->
---
 src/common/utils/ai/modelCapabilities.test.ts |  6 ++++++
 src/common/utils/tokens/modelStats.test.ts    |  7 +++++++
 src/common/utils/tokens/models-extra.ts       | 17 +++++++++++++++++
 3 files changed, 30 insertions(+)

diff --git a/src/common/utils/ai/modelCapabilities.test.ts b/src/common/utils/ai/modelCapabilities.test.ts
index 95287de89a..0408503d68 100644
--- a/src/common/utils/ai/modelCapabilities.test.ts
+++ b/src/common/utils/ai/modelCapabilities.test.ts
@@ -23,6 +23,12 @@ describe("getModelCapabilities", () => {
     expect(caps).not.toBeNull();
   });
 
+  it("keeps GPT-5.3-Codex-Spark as text-only via models-extra override", () => {
+    const caps = getModelCapabilities("openai:gpt-5.3-codex-spark");
+    expect(caps).not.toBeNull();
+    expect(caps?.supportsVision).toBe(false);
+  });
+
   it("returns maxPdfSizeMb when present in model metadata", () => {
     const caps = getModelCapabilities("google:gemini-1.5-flash");
     expect(caps).not.toBeNull();
diff --git a/src/common/utils/tokens/modelStats.test.ts b/src/common/utils/tokens/modelStats.test.ts
index 7ae128aa1c..f17fbd3091 100644
--- a/src/common/utils/tokens/modelStats.test.ts
+++ b/src/common/utils/tokens/modelStats.test.ts
@@ -24,6 +24,13 @@ describe("getModelStats", () => {
       expect(stats?.input_cost_per_token).toBe(0.000021);
     });
 
+    test("should find GPT-5.3-Codex-Spark in models-extra.ts", () => {
+      const stats = getModelStats("openai:gpt-5.3-codex-spark");
+      expect(stats).not.toBeNull();
+      expect(stats?.max_input_tokens).toBe(128000);
+      expect(stats?.max_output_tokens).toBe(128000);
+    });
+
     test("models-extra.ts should override models.json", () => {
       // gpt-5.2-codex exists in both files - models-extra.ts has correct 272k, models.json has incorrect 400k
       const stats = getModelStats("openai:gpt-5.2-codex");
diff --git a/src/common/utils/tokens/models-extra.ts b/src/common/utils/tokens/models-extra.ts
index 91391b5a95..64c1a994f0 100644
--- a/src/common/utils/tokens/models-extra.ts
+++ b/src/common/utils/tokens/models-extra.ts
@@ -109,6 +109,23 @@ export const modelsExtra: Record<string, ModelData> = {
     supports_response_schema: true,
   },
 
+  // GPT-5.3-Codex-Spark - Released February 13, 2026
+  // OpenAI says this is a text-only Codex variant with a 128k context window.
+  // API pricing is not published yet, so we temporarily mirror GPT-5.3-Codex pricing.
+  "gpt-5.3-codex-spark": {
+    max_input_tokens: 128000,
+    max_output_tokens: 128000,
+    input_cost_per_token: 0.00000175, // $1.75 per million input tokens (temporary)
+    output_cost_per_token: 0.000014, // $14 per million output tokens (temporary)
+    cache_read_input_token_cost: 0.000000175, // $0.175 per million cached input tokens (temporary)
+    litellm_provider: "openai",
+    mode: "responses",
+    supports_function_calling: true,
+    supports_vision: false,
+    supports_reasoning: true,
+    supports_response_schema: true,
+  },
+
   // GPT-5.2 Pro - Released December 11, 2025
   // $21/M input, $168/M output
   // Supports medium, high, xhigh reasoning levels