diff --git a/.changeset/add-litellm-adapter.md b/.changeset/add-litellm-adapter.md
new file mode 100644
index 000000000..cd27642cb
--- /dev/null
+++ b/.changeset/add-litellm-adapter.md
@@ -0,0 +1,7 @@
+---
+'@tanstack/ai-litellm': minor
+---
+
+feat: add LiteLLM AI gateway adapter
+
+New `@tanstack/ai-litellm` package that provides a tree-shakeable text adapter for the LiteLLM proxy. Extends `OpenAIBaseChatCompletionsTextAdapter` (same pattern as `ai-groq` and `ai-grok`), giving access to 100+ LLM providers through a single adapter.
diff --git a/packages/ai-litellm/package.json b/packages/ai-litellm/package.json
new file mode 100644
index 000000000..bda415bc5
--- /dev/null
+++ b/packages/ai-litellm/package.json
@@ -0,0 +1,60 @@
+{
+  "name": "@tanstack/ai-litellm",
+  "version": "0.0.1",
+  "type": "module",
+  "description": "LiteLLM AI gateway adapter for TanStack AI. Access 100+ LLM providers through a single proxy.",
+  "author": "",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/TanStack/ai.git",
+    "directory": "packages/ai-litellm"
+  },
+  "module": "./dist/esm/index.js",
+  "types": "./dist/esm/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/esm/index.d.ts",
+      "import": "./dist/esm/index.js"
+    }
+  },
+  "files": [
+    "dist",
+    "src"
+  ],
+  "scripts": {
+    "build": "vite build",
+    "clean": "premove ./build ./dist",
+    "lint:fix": "eslint ./src --fix",
+    "test:build": "publint --strict",
+    "test:eslint": "eslint ./src",
+    "test:lib": "vitest run",
+    "test:lib:dev": "pnpm test:lib --watch",
+    "test:types": "tsc"
+  },
+  "keywords": [
+    "ai",
+    "ai-sdk",
+    "typescript",
+    "tanstack",
+    "litellm",
+    "adapter",
+    "llm",
+    "gateway",
+    "multi-provider",
+    "proxy"
+  ],
+  "devDependencies": {
+    "@vitest/coverage-v8": "4.0.14",
+    "vite": "^7.3.3"
+  },
+  "peerDependencies": {
+    "@tanstack/ai": "workspace:^",
+    "zod": "^4.0.0"
+  },
+  "dependencies": {
+    "@tanstack/ai-utils": "workspace:*",
+    "@tanstack/openai-base": "workspace:*",
+    "openai": "^6.41.0"
+  }
+}
diff --git a/packages/ai-litellm/src/adapters/text.ts b/packages/ai-litellm/src/adapters/text.ts
new file mode 100644
index 000000000..600cdd271
--- /dev/null
+++ b/packages/ai-litellm/src/adapters/text.ts
@@ -0,0 +1,69 @@
+import OpenAI from 'openai'
+import { OpenAIBaseChatCompletionsTextAdapter } from '@tanstack/openai-base'
+import { getLiteLLMApiKeyFromEnv, withLiteLLMDefaults } from '../utils/client'
+import type { LiteLLMClientConfig } from '../utils'
+
+/**
+ * Configuration for the LiteLLM text adapter.
+ */
+export interface LiteLLMTextConfig extends LiteLLMClientConfig {}
+
+/**
+ * LiteLLM Text (Chat) Adapter
+ *
+ * Tree-shakeable adapter for LiteLLM AI gateway. LiteLLM exposes an
+ * OpenAI-compatible Chat Completions endpoint, so we drive it with the
+ * OpenAI SDK via a baseURL override (the same pattern as ai-groq and
+ * ai-grok).
+ *
+ * LiteLLM supports 100+ providers (OpenAI, Anthropic, Google, Azure,
+ * AWS Bedrock, Ollama, Groq, Mistral, and more) through a single proxy.
+ * The model string determines the provider routing, e.g.
+ * "anthropic/claude-sonnet-4-6" or "openai/gpt-4o".
+ */
+export class LiteLLMTextAdapter extends OpenAIBaseChatCompletionsTextAdapter<
+  string,
+  Record<string, any>,
+  readonly [],
+  Record<string, never>,
+  readonly []
+> {
+  override readonly kind = 'text' as const
+  override readonly name = 'litellm' as const
+
+  constructor(config: LiteLLMTextConfig, model: string) {
+    super(model, 'litellm', new OpenAI(withLiteLLMDefaults(config)))
+  }
+}
+
+/**
+ * Creates a LiteLLM text adapter with explicit API key.
+ *
+ * @example
+ * ```typescript
+ * const adapter = createLitellmText('anthropic/claude-sonnet-4-6', 'sk-...');
+ * ```
+ */
+export function createLitellmText(
+  model: string,
+  apiKey: string,
+  config?: Omit<LiteLLMTextConfig, 'apiKey'>,
+): LiteLLMTextAdapter {
+  return new LiteLLMTextAdapter({ apiKey, ...config }, model)
+}
+
+/**
+ * Creates a LiteLLM text adapter with API key from `LITELLM_API_KEY`.
+ *
+ * @example
+ * ```typescript
+ * const adapter = litellmText('anthropic/claude-sonnet-4-6');
+ * ```
+ */
+export function litellmText(
+  model: string,
+  config?: Omit<LiteLLMTextConfig, 'apiKey'>,
+): LiteLLMTextAdapter {
+  const apiKey = getLiteLLMApiKeyFromEnv()
+  return createLitellmText(model, apiKey, config)
+}
diff --git a/packages/ai-litellm/src/index.ts b/packages/ai-litellm/src/index.ts
new file mode 100644
index 000000000..850de7d16
--- /dev/null
+++ b/packages/ai-litellm/src/index.ts
@@ -0,0 +1,22 @@
+/**
+ * @module @tanstack/ai-litellm
+ *
+ * LiteLLM AI gateway adapter for TanStack AI.
+ * Provides a tree-shakeable adapter for LiteLLM's OpenAI-compatible proxy,
+ * giving access to 100+ LLM providers through a single interface.
+ */
+
+// Text (Chat) adapter
+export {
+  LiteLLMTextAdapter,
+  createLitellmText,
+  litellmText,
+  type LiteLLMTextConfig,
+} from './adapters/text'
+
+// Utilities
+export {
+  getLiteLLMApiKeyFromEnv,
+  withLiteLLMDefaults,
+  type LiteLLMClientConfig,
+} from './utils'
diff --git a/packages/ai-litellm/src/utils/client.ts b/packages/ai-litellm/src/utils/client.ts
new file mode 100644
index 000000000..61a674a66
--- /dev/null
+++ b/packages/ai-litellm/src/utils/client.ts
@@ -0,0 +1,35 @@
+import { getApiKeyFromEnv } from '@tanstack/ai-utils'
+import type { ClientOptions } from 'openai'
+
+export interface LiteLLMClientConfig extends Omit<ClientOptions, 'apiKey'> {
+  apiKey?: string
+}
+
+const DEFAULT_LITELLM_BASE_URL = 'http://localhost:4000/v1'
+
+/**
+ * Gets LiteLLM API key from environment variables.
+ * @throws Error if LITELLM_API_KEY is not found
+ */
+export function getLiteLLMApiKeyFromEnv(): string {
+  try {
+    return getApiKeyFromEnv('LITELLM_API_KEY')
+  } catch {
+    throw new Error(
+      'LITELLM_API_KEY is required. Please set it in your environment variables or use createLitellmText() with an explicit API key.',
+    )
+  }
+}
+
+/**
+ * Returns an OpenAI client config pointing at the LiteLLM proxy.
+ * Defaults to http://localhost:4000/v1 when no baseURL is provided.
+ */
+export function withLiteLLMDefaults(
+  config: LiteLLMClientConfig,
+): LiteLLMClientConfig {
+  return {
+    ...config,
+    baseURL: config.baseURL || DEFAULT_LITELLM_BASE_URL,
+  }
+}
diff --git a/packages/ai-litellm/src/utils/index.ts b/packages/ai-litellm/src/utils/index.ts
new file mode 100644
index 000000000..fb34ea388
--- /dev/null
+++ b/packages/ai-litellm/src/utils/index.ts
@@ -0,0 +1,5 @@
+export {
+  getLiteLLMApiKeyFromEnv,
+  withLiteLLMDefaults,
+  type LiteLLMClientConfig,
+} from './client'
diff --git a/packages/ai-litellm/tsconfig.json b/packages/ai-litellm/tsconfig.json
new file mode 100644
index 000000000..c38689f4e
--- /dev/null
+++ b/packages/ai-litellm/tsconfig.json
@@ -0,0 +1,8 @@
+{
+  "extends": "../../tsconfig.base.json",
+  "compilerOptions": {
+    "outDir": "dist"
+  },
+  "include": ["src", "tests"],
+  "exclude": ["node_modules", "dist"]
+}
diff --git a/packages/ai-litellm/vite.config.ts b/packages/ai-litellm/vite.config.ts
new file mode 100644
index 000000000..77bcc2e60
--- /dev/null
+++ b/packages/ai-litellm/vite.config.ts
@@ -0,0 +1,36 @@
+import { defineConfig, mergeConfig } from 'vitest/config'
+import { tanstackViteConfig } from '@tanstack/vite-config'
+import packageJson from './package.json'
+
+const config = defineConfig({
+  test: {
+    name: packageJson.name,
+    dir: './',
+    watch: false,
+    globals: true,
+    environment: 'node',
+    include: ['tests/**/*.test.ts'],
+    coverage: {
+      provider: 'v8',
+      reporter: ['text', 'json', 'html', 'lcov'],
+      exclude: [
+        'node_modules/',
+        'dist/',
+        'tests/',
+        '**/*.test.ts',
+        '**/*.config.ts',
+        '**/types.ts',
+      ],
+      include: ['src/**/*.ts'],
+    },
+  },
+})
+
+export default mergeConfig(
+  config,
+  tanstackViteConfig({
+    entry: ['./src/index.ts'],
+    srcDir: './src',
+    cjs: false,
+  }),
+)
diff --git a/testing/e2e/src/lib/feature-support.ts b/testing/e2e/src/lib/feature-support.ts
index 6d6b950bd..65890433c 100644
--- a/testing/e2e/src/lib/feature-support.ts
+++ b/testing/e2e/src/lib/feature-support.ts
@@ -17,6 +17,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   'one-shot-text': new Set([
     'openai',
@@ -27,6 +28,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   reasoning: new Set(['openai', 'anthropic', 'gemini']),
   'multi-turn': new Set([
@@ -38,6 +40,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   'tool-calling': new Set([
     'openai',
@@ -48,6 +51,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   'parallel-tool-calls': new Set([
     'openai',
@@ -57,6 +61,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   // Gemini excluded: approval flow timing issues with Gemini's streaming format
   'tool-approval': new Set([
@@ -67,6 +72,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   // Ollama excluded: aimock doesn't support content+toolCalls for /api/chat format
   'text-tool-text': new Set([
@@ -77,6 +83,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   'structured-output': new Set([
     'openai',
@@ -87,6 +94,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   // Streaming structured output: only providers with native streaming JSON
   // schema support are listed here. Other providers fall back to the
@@ -98,6 +106,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   // Multi-turn structured output: every turn produces its own typed
   // `structured-output` part on the assistant message, and historical
@@ -125,6 +134,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   'agentic-structured': new Set([
     'openai',
@@ -135,6 +145,7 @@ export const matrix: Record<Feature, Set<Provider>> = {
     'grok',
     'openrouter',
     'openai-compatible',
+    'litellm',
   ]),
   // Native-combined-mode adapters only. Each provider's default test model
   // (or per-feature override in `features.ts`) must opt into combined mode
diff --git a/testing/e2e/src/lib/types.ts b/testing/e2e/src/lib/types.ts
index 018e7744f..314d5a267 100644
--- a/testing/e2e/src/lib/types.ts
+++ b/testing/e2e/src/lib/types.ts
@@ -10,6 +10,7 @@ export type Provider =
   | 'openrouter'
   | 'openrouter-responses'
   | 'openai-compatible'
+  | 'litellm'
   | 'elevenlabs'
 
 export type Feature =
@@ -48,6 +49,7 @@ export const ALL_PROVIDERS: Provider[] = [
   'openrouter',
   'openrouter-responses',
   'openai-compatible',
+  'litellm',
   'elevenlabs',
 ]
 
diff --git a/testing/e2e/tests/test-matrix.ts b/testing/e2e/tests/test-matrix.ts
index 58a166a3a..cd0cfe956 100644
--- a/testing/e2e/tests/test-matrix.ts
+++ b/testing/e2e/tests/test-matrix.ts
@@ -23,6 +23,7 @@ export const providers: Provider[] = [
   'openrouter',
   'openrouter-responses',
   'openai-compatible',
+  'litellm',
   'elevenlabs',
 ]