AlemTuzlak · Jan 5, 2026 · Dec 26, 2025 · Dec 26, 2025 · Dec 26, 2025 · Dec 26, 2025
diff --git a/.changeset/full-stars-drop.md b/.changeset/full-stars-drop.md
@@ -0,0 +1,7 @@
+---
+'@tanstack/ai-openai': patch
+'@tanstack/ai': patch
+'@tanstack/ai-solid': patch
+---
+
+Fix up model names for OpenAI and release the new response API's
diff --git a/.changeset/mighty-gifts-enter.md b/.changeset/mighty-gifts-enter.md
@@ -0,0 +1,14 @@
+---
+'@tanstack/react-ai-devtools': patch
+'@tanstack/solid-ai-devtools': patch
+'@tanstack/ai-devtools-core': patch
+'@tanstack/ai-react-ui': patch
+'@tanstack/ai-client': patch
+'@tanstack/ai-gemini': patch
+'@tanstack/ai-ollama': patch
+'@tanstack/ai-openai': patch
+'@tanstack/ai-react': patch
+'@tanstack/ai': patch
+---
+
+fix up readmes
diff --git a/README.md b/README.md
@@ -72,31 +72,6 @@ for await (const chunk of result) {
 
 Available adapters: `openaiText`, `openaiEmbed`, `openaiSummarize`, `anthropicText`, `geminiText`, `ollamaText`, and more.
 
-## Bonus: TanStack Start Integration
-
-TanStack AI works with **any** framework (Next.js, Express, Remix, etc.).
-
-**With TanStack Start**, you get a bonus: share implementations between AI tools and server functions with `createServerFnTool`:
-
-```typescript
-import { createServerFnTool } from '@tanstack/ai-react'
-
-// Define once, get AI tool AND server function (TanStack Start only)
-const getProducts = createServerFnTool({
-  name: 'getProducts',
-  inputSchema: z.object({ query: z.string() }),
-  execute: async ({ query }) => db.products.search(query),
-})
-
-// Use in AI chat
-chat({ tools: [getProducts.server] })
-
-// Call directly from components (no API endpoint needed!)
-const products = await getProducts.serverFn({ query: 'laptop' })
-```
-
-No duplicate logic, full type safety, automatic validation. The `serverFn` feature requires TanStack Start. See [docs](https://tanstack.com/ai) for details.
-
 ## Get Involved
 
 - We welcome issues and pull requests!

diff --git a/packages/typescript/ai-client/README.md b/packages/typescript/ai-client/README.md
@@ -72,31 +72,6 @@ for await (const chunk of result) {
 
 Available adapters: `openaiText`, `openaiEmbed`, `openaiSummarize`, `anthropicText`, `geminiText`, `ollamaText`, and more.
 
-## Bonus: TanStack Start Integration
-
-TanStack AI works with **any** framework (Next.js, Express, Remix, etc.).
-
-**With TanStack Start**, you get a bonus: share implementations between AI tools and server functions with `createServerFnTool`:
-
-```typescript
-import { createServerFnTool } from '@tanstack/ai-react'
-
-// Define once, get AI tool AND server function (TanStack Start only)
-const getProducts = createServerFnTool({
-  name: 'getProducts',
-  inputSchema: z.object({ query: z.string() }),
-  execute: async ({ query }) => db.products.search(query),
-})
-
-// Use in AI chat
-chat({ tools: [getProducts.server] })
-
-// Call directly from components (no API endpoint needed!)
-const products = await getProducts.serverFn({ query: 'laptop' })
-```
-
-No duplicate logic, full type safety, automatic validation. The `serverFn` feature requires TanStack Start. See [docs](https://tanstack.com/ai) for details.
-
 ## Get Involved
 
 - We welcome issues and pull requests!

diff --git a/packages/typescript/ai-devtools/README.md b/packages/typescript/ai-devtools/README.md
@@ -72,31 +72,6 @@ for await (const chunk of result) {
 
 Available adapters: `openaiText`, `openaiEmbed`, `openaiSummarize`, `anthropicText`, `geminiText`, `ollamaText`, and more.
 
-## Bonus: TanStack Start Integration
-
-TanStack AI works with **any** framework (Next.js, Express, Remix, etc.).
-
-**With TanStack Start**, you get a bonus: share implementations between AI tools and server functions with `createServerFnTool`:
-
-```typescript
-import { createServerFnTool } from '@tanstack/ai-react'
-
-// Define once, get AI tool AND server function (TanStack Start only)
-const getProducts = createServerFnTool({
-  name: 'getProducts',
-  inputSchema: z.object({ query: z.string() }),
-  execute: async ({ query }) => db.products.search(query),
-})
-
-// Use in AI chat
-chat({ tools: [getProducts.server] })
-
-// Call directly from components (no API endpoint needed!)
-const products = await getProducts.serverFn({ query: 'laptop' })
-```
-
-No duplicate logic, full type safety, automatic validation. The `serverFn` feature requires TanStack Start. See [docs](https://tanstack.com/ai) for details.
-
 ## Get Involved
 
 - We welcome issues and pull requests!

diff --git a/packages/typescript/ai-gemini/README.md b/packages/typescript/ai-gemini/README.md
@@ -72,31 +72,6 @@ for await (const chunk of result) {
 
 Available adapters: `openaiText`, `openaiEmbed`, `openaiSummarize`, `anthropicText`, `geminiText`, `ollamaText`, and more.
 
-## Bonus: TanStack Start Integration
-
-TanStack AI works with **any** framework (Next.js, Express, Remix, etc.).
-
-**With TanStack Start**, you get a bonus: share implementations between AI tools and server functions with `createServerFnTool`:
-
-```typescript
-import { createServerFnTool } from '@tanstack/ai-react'
-
-// Define once, get AI tool AND server function (TanStack Start only)
-const getProducts = createServerFnTool({
-  name: 'getProducts',
-  inputSchema: z.object({ query: z.string() }),
-  execute: async ({ query }) => db.products.search(query),
-})
-
-// Use in AI chat
-chat({ tools: [getProducts.server] })
-
-// Call directly from components (no API endpoint needed!)
-const products = await getProducts.serverFn({ query: 'laptop' })
-```
-
-No duplicate logic, full type safety, automatic validation. The `serverFn` feature requires TanStack Start. See [docs](https://tanstack.com/ai) for details.
-
 ## Get Involved
 
 - We welcome issues and pull requests!

diff --git a/packages/typescript/ai-ollama/README.md b/packages/typescript/ai-ollama/README.md
@@ -72,31 +72,6 @@ for await (const chunk of result) {
 
 Available adapters: `openaiText`, `openaiEmbed`, `openaiSummarize`, `anthropicText`, `geminiText`, `ollamaText`, and more.
 
-## Bonus: TanStack Start Integration
-
-TanStack AI works with **any** framework (Next.js, Express, Remix, etc.).
-
-**With TanStack Start**, you get a bonus: share implementations between AI tools and server functions with `createServerFnTool`:
-
-```typescript
-import { createServerFnTool } from '@tanstack/ai-react'
-
-// Define once, get AI tool AND server function (TanStack Start only)
-const getProducts = createServerFnTool({
-  name: 'getProducts',
-  inputSchema: z.object({ query: z.string() }),
-  execute: async ({ query }) => db.products.search(query),
-})
-
-// Use in AI chat
-chat({ tools: [getProducts.server] })
-
-// Call directly from components (no API endpoint needed!)
-const products = await getProducts.serverFn({ query: 'laptop' })
-```
-
-No duplicate logic, full type safety, automatic validation. The `serverFn` feature requires TanStack Start. See [docs](https://tanstack.com/ai) for details.
-
 ## Get Involved
 
 - We welcome issues and pull requests!

diff --git a/packages/typescript/ai-openai/README.md b/packages/typescript/ai-openai/README.md
@@ -72,31 +72,6 @@ for await (const chunk of result) {
 
 Available adapters: `openaiText`, `openaiEmbed`, `openaiSummarize`, `anthropicText`, `geminiText`, `ollamaText`, and more.
 
-## Bonus: TanStack Start Integration
-
-TanStack AI works with **any** framework (Next.js, Express, Remix, etc.).
-
-**With TanStack Start**, you get a bonus: share implementations between AI tools and server functions with `createServerFnTool`:
-
-```typescript
-import { createServerFnTool } from '@tanstack/ai-react'
-
-// Define once, get AI tool AND server function (TanStack Start only)
-const getProducts = createServerFnTool({
-  name: 'getProducts',
-  inputSchema: z.object({ query: z.string() }),
-  execute: async ({ query }) => db.products.search(query),
-})
-
-// Use in AI chat
-chat({ tools: [getProducts.server] })
-
-// Call directly from components (no API endpoint needed!)
-const products = await getProducts.serverFn({ query: 'laptop' })
-```
-
-No duplicate logic, full type safety, automatic validation. The `serverFn` feature requires TanStack Start. See [docs](https://tanstack.com/ai) for details.
-
 ## Get Involved
 
 - We welcome issues and pull requests!

diff --git a/packages/typescript/ai-openai/src/model-meta.ts b/packages/typescript/ai-openai/src/model-meta.ts
@@ -133,7 +133,7 @@ const GPT5_2_PRO = {
 >
 
 const GPT5_2_CHAT = {
-  name: 'gpt-5.2-chat',
+  name: 'gpt-5.2-chat-latest',
   context_window: 128_000,
   max_output_tokens: 16_384,
   knowledge_cutoff: '2025-08-31',
@@ -1271,7 +1271,7 @@ const GPT_4_TURBO = {
 >
 
 const CHATGPT_40 = {
-  name: 'chatgpt-4.0',
+  name: 'chatgpt-4o-latest',
   context_window: 128_000,
   max_output_tokens: 4_096,
   knowledge_cutoff: '2023-10-01',
@@ -1539,7 +1539,7 @@ const GPT_4O_TRANSCRIBE_DIARIZE = {
 > */
 
 const GPT_5_1_CHAT = {
-  name: 'gpt-5.1-chat',
+  name: 'gpt-5.1-chat-latest',
   context_window: 128_000,
   max_output_tokens: 16_384,
   knowledge_cutoff: '2024-09-30',
@@ -1568,7 +1568,7 @@ const GPT_5_1_CHAT = {
 >
 
 const GPT_5_CHAT = {
-  name: 'gpt-5-chat',
+  name: 'gpt-5-chat-latest',
   context_window: 128_000,
   max_output_tokens: 16_384,
   knowledge_cutoff: '2024-09-30',

diff --git a/packages/typescript/ai-openai/src/utils/schema-converter.ts b/packages/typescript/ai-openai/src/utils/schema-converter.ts
@@ -61,7 +61,7 @@ export function makeOpenAIStructuredOutputCompatible(
       const prop = properties[propName]
       const wasOptional = !originalRequired.includes(propName)
 
-      // Recursively transform nested objects/arrays
+      // Recursively transform nested objects/arrays/unions
       if (prop.type === 'object' && prop.properties) {
         properties[propName] = makeOpenAIStructuredOutputCompatible(
           prop,
@@ -75,6 +75,12 @@ export function makeOpenAIStructuredOutputCompatible(
             prop.items.required || [],
           ),
         }
+      } else if (prop.anyOf) {
+        // Handle anyOf/oneOf at property level (union types)
+        properties[propName] = makeOpenAIStructuredOutputCompatible(
+          prop,
+          prop.required || [],
+        )
       } else if (wasOptional) {
         // Make optional fields nullable by adding null to the type
         if (prop.type && !Array.isArray(prop.type)) {
@@ -106,5 +112,12 @@ export function makeOpenAIStructuredOutputCompatible(
     )
   }
 
+  // Handle anyOf (union types) - each variant needs to be transformed
+  if (result.anyOf && Array.isArray(result.anyOf)) {
+    result.anyOf = result.anyOf.map((variant) =>
+      makeOpenAIStructuredOutputCompatible(variant, variant.required || []),
+    )
+  }
+
   return result
 }