@neon/ai-sdk-provider 0.7.2 → 0.8.0

package/{LICENSE.md → LICENSE}

@@ -1,6 +1,5 @@
-# Apache License 2.0
 
-                                Apache License
+                                 Apache License
                            Version 2.0, January 2004
                         http://www.apache.org/licenses/
 
@@ -175,4 +174,29 @@
       incurred by, or claims asserted against, such Contributor by reason
       of your accepting any such warranty or additional liability.
 
-   END OF TERMS AND CONDITIONS
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

package/README.md

@@ -2,9 +2,9 @@
 
 Community [Vercel AI SDK](https://ai-sdk.dev) provider for the [Neon](https://neon.com) AI Gateway. Supports **AI SDK v6 and v7** (`ai@^6` or `ai@^7`).
 
-The Neon AI Gateway is **branch-scoped**: each Neon project branch gets its own gateway host, and a platform token authorizes requests for that branch. This provider routes each model to the best gateway endpoint (Anthropic → native Messages, OpenAI → native Responses incl. **Codex**, everything else → unified OpenAI-compatible MLflow endpoint), so a single `neon('claude-...')` call reaches the whole catalog.
+The Neon AI Gateway is **branch-scoped**: each Neon project branch gets its own gateway host, and a platform token authorizes requests for that branch. Use the same `neon(modelId)` API across the branch's model catalog; the provider selects Anthropic Messages, OpenAI Responses, or Chat Completions for each model.
 
-Model ids use the canonical Neon (unprefixed) form — `claude-sonnet-4-6`, `gpt-5`, `gemini-2-5-flash` — matching the [`neon` provider on models.dev](https://models.dev). The typed catalog mirrors that provider exactly (kept in sync by a scheduled drift check), plus a few extra gateway-served ids that models.dev doesn't list yet (e.g. Codex, Llama, Qwen). Any other id — including the legacy `databricks-` prefixed form (`databricks-claude-sonnet-4-6`) — is still accepted as a plain string, so existing code keeps working.
+Use canonical model ids such as `gpt-5-mini`, `llama-4-maverick`, and `gemini-3-flash`, matching the [`neon` provider on models.dev](https://models.dev). The typed catalog includes the known model ids, and arbitrary strings are accepted so newly available models work before the types update.
 
 ## Install
 
@@ -31,7 +31,7 @@ import { generateText } from "ai";
 
 // Reads NEON_AI_GATEWAY_BASE_URL + NEON_AI_GATEWAY_TOKEN from the environment.
 const { text } = await generateText({
-  model: neon("claude-haiku-4-5"), // or 'gpt-5-3-codex', etc.
+  model: neon("gpt-5-mini"), // or "llama-4-maverick", "gemini-3-flash", …
   prompt: "Summarize Postgres for me.",
 });
 ```
@@ -51,17 +51,56 @@ const neon = createNeon({
 
 | Model family | Endpoint | Why |
 | --- | --- | --- |
-| Anthropic (`claude-*`) | native Messages API | streaming structured output + native reasoning |
-| OpenAI (`gpt-*`, `*-codex`) | native Responses API | Codex (native-only), native reasoning, image-gen tool |
-| Everything else (Gemini, Llama, Qwen, gpt-oss, ...) | unified MLflow endpoint | broad coverage; Gemini is here because its native endpoint does not support streaming |
+| Anthropic (`claude-*`) | Messages API | Anthropic tools, structured output, and reasoning |
+| OpenAI (`gpt-*`, `*-codex`) | Responses API | Codex, reasoning, and built-in tools such as image generation |
+| Everything else (Gemini, Llama, Qwen, gpt-oss, ...) | Chat Completions API | one streaming interface across the remaining model families |
 
-Routing matches on the model id, so both the canonical (`gpt-5`) and the legacy `databricks-`-prefixed (`databricks-gpt-5`) forms route identically.
+Routing matches on the model id.
 
 ## Capabilities
 
-Verified across Anthropic, OpenAI (incl. Codex), Google, and Meta models: `generateText` / `streamText` (text, system prompts, multi-turn), tool calling (single and multi-step, generate and stream), `generateObject` / `streamObject`, and image (vision) input.
+`generateText` and `streamText` work with any model available to your branch. `generateObject`, `streamObject`, and single- or multi-step tool calls work with OpenAI (including Codex), Meta, Alibaba, Zhipu AI, and Thinking Machines models. Gemini currently supports `generateText` and `streamText`; structured output and multi-step tools are not supported. Vision input works on models that accept images.
 
-For MLflow-routed models, the provider detects the model family and drops parameters a backend rejects (e.g. penalties/`seed` for Llama, `reasoningEffort` for Gemini) with an AI SDK warning (`result.warnings`) instead of failing the request.
+Claude models use the Messages API. On both that route and Chat Completions the provider removes call options the gateway rejects and reports each one in `result.warnings` instead of failing the request — see [Dropped call options](#dropped-call-options). Unsupported Responses API storage options throw before a request is sent — see [Errors](#errors).
+
+Which ids your branch serves is account-specific during the beta; `GET $NEON_AI_GATEWAY_BASE_URL/v1/models` is the authoritative list. `NeonChatModelId` accepts any string, so an id this package has not caught up with still works — and, equally, an id the gateway has retired still type-checks and fails at the gateway instead.
+
+### Dropped call options
+
+Upstream backends behind the gateway accept different subsets of the OpenAI-style parameters the AI SDK emits, and sending one an upstream rejects is a hard `400`. The provider drops those before the request and records a warning in `result.warnings`, so a call succeeds instead of failing on a parameter you passed in good faith.
+
+These rules apply on the Anthropic Messages and Chat Completions routes. The Responses route (every `gpt-*` id) relies on the upstream OpenAI model's own stripping instead, so its behaviour is the AI SDK's rather than described here — but `getNeonModelCapabilities` still answers for those ids.
+
+| Models | Dropped |
+| --- | --- |
+| Claude 4.7 and newer (`claude-opus-4-7`, `claude-opus-4-8`, `claude-opus-5`, `claude-sonnet-5`, `claude-fable-5`) | `temperature`, `topP` |
+| All Claude | `frequencyPenalty`, `presencePenalty`, `seed`, `reasoningEffort`, and `topP` when you set both `temperature` and `topP` (Anthropic accepts only one) |
+| `gpt-oss` | `frequencyPenalty`, `presencePenalty`, `seed`, `stopSequences` |
+| Qwen, Gemma | `frequencyPenalty`, `presencePenalty`, `seed` |
+| `glm-5-2`, `inkling` | `frequencyPenalty`, `presencePenalty` |
+| Meta Llama | `frequencyPenalty`, `presencePenalty`, `seed` |
+| Gemini | `reasoningEffort` |
+
+**`temperature` on a Claude 5 model therefore has no effect.** That is the gateway's behaviour, not a provider choice; sending it returns `does not support the temperature parameter`. Steer those models with Anthropic's own effort control instead — note that this is `effort`, not the OpenAI-style `reasoningEffort`, which every Claude id drops:
+
+```typescript
+await generateText({
+  model: neon('claude-opus-5'),
+  prompt: 'Plan a schema migration.',
+  providerOptions: { anthropic: { effort: 'high' } }, // low | medium | high | xhigh | max
+});
+```
+
+Query the rules directly rather than hardcoding them:
+
+```typescript
+import { getNeonModelCapabilities } from '@neon/ai-sdk-provider';
+
+getNeonModelCapabilities('claude-opus-5').supportsTemperature; // false
+getNeonModelCapabilities('glm-5-2').supportsPenalties; // false
+```
+
+Gemini is the only family on the unified endpoint that accepts penalties. A model none of these rules match is left untouched, so a brand-new id gets the gateway's own error rather than a guess.
 
 ## Image generation
 
@@ -85,11 +124,90 @@ for await (const part of result.fullStream) {
 }
 ```
 
+### Edit a generated image
+
+The gateway does not support stored Responses API items (`store: true`, `previousResponseId`, or `conversation`), which the Responses API uses to reuse tool results across turns. For image edits, pass the returned bytes back as image input; otherwise the model does not receive the original image and generates a new one.
+
+```ts
+const first = await generateText({
+  model: neon("gpt-5-mini"),
+  prompt: "Generate an image of a plain red square.",
+  tools: { image_generation: neon.tools.imageGeneration({ outputFormat: "jpeg" }) },
+});
+
+const generated = first.steps
+  .flatMap((step) => step.toolResults)
+  .find((result) => result.toolName === "image_generation");
+const output = generated?.output;
+if (
+  !output ||
+  typeof output !== "object" ||
+  !("result" in output) ||
+  typeof output.result !== "string"
+) {
+  throw new Error("Image generation did not return image bytes.");
+}
+
+// Pass it back as image content, not as conversation history.
+await generateText({
+  model: neon("gpt-5-mini"),
+  tools: { image_generation: neon.tools.imageGeneration({ outputFormat: "jpeg" }) },
+  messages: [
+    {
+      role: "user",
+      content: [
+        { type: "image", image: output.result, mediaType: "image/jpeg" },
+        { type: "text", text: "Make that same square blue instead." },
+      ],
+    },
+  ],
+});
+```
+
+The same stateless limitation applies to other built-in Responses tools such as web search and code interpreter.
+
+## Errors
+
+A failed call rejects with the AI SDK's `APICallError`, and `error.message` carries the gateway's own reason rather than the bare HTTP status line:
+
+```ts
+import { APICallError, generateText } from "ai";
+
+try {
+  await generateText({ model: neon("gpt-5-mini"), maxOutputTokens: 1, prompt });
+} catch (error) {
+  if (APICallError.isInstance(error)) {
+    error.message;      // "Invalid 'max_output_tokens': integer below minimum
+                        //  value. Expected a value >= 16, but got 1 instead."
+    error.responseBody; // the gateway's original body, including its error_code
+  }
+}
+```
+
+The provider normalizes JSON error responses so the gateway's reason lands on `error.message`. Errors delivered inside an open stream and non-JSON responses keep the HTTP status line; inspect `error.responseBody` for their payload.
+
+Requests that cannot work are refused before they leave, with `UnsupportedFunctionalityError`:
+
+| `providerOptions.openai` | Why |
+| --- | --- |
+| `store: true` or `store: null` | The Responses route is stateless and never persists items. `null` is not "omitted" — the AI SDK reads it as `true`. |
+| `previousResponseId`, `conversation` | Nothing is stored for them to refer to. Send the full message history instead. |
+
+Provider options are namespaced per route, which decides where the SDK reads them from:
+
+| Route | Models | Namespace |
+| --- | --- | --- |
+| Responses | `gpt-*` (incl. Codex) | `openai` |
+| Chat Completions | Gemini, Llama, Qwen, gpt-oss, … | `neon` |
+| Anthropic Messages | `claude-*` | `anthropic` |
+
+So `store` is only refused on the Responses route; the same option is ignored elsewhere.
+
 ## Limitations
 
-- `generateImage()` and embeddings (`embed` / `embedMany`) are not offered by the gateway and throw `NoSuchModelError`.
-- `gpt-oss-*` models return a non-standard ("harmony") response shape on the unified endpoint (`message.content` as an array of reasoning/text parts instead of a string). The provider normalizes this to the OpenAI Chat Completions contract (string `content` + `reasoning_content`) so `generateText`/`streamText` work and reasoning is surfaced. See neondatabase/neon-pkgs#308.
-- OpenAI Responses multi-turn tool flows (`generateText` + `stepCountIs`) can return 502 from the gateway; tool calling is covered on Anthropic/Google/Meta in e2e.
+- The gateway does not offer image or embedding model endpoints, so `generateImage()`, `embed()`, and `embedMany()` throw `NoSuchModelError`. Image generation is available through the Responses API's built-in `image_generation` tool with `neon.tools.imageGeneration()`.
+- Results from provider-executed tools (`neon.tools.imageGeneration`, and the other Responses built-ins) are not replayed to the gateway on a later step — see [Edit a generated image](#edit-a-generated-image).
+- The Responses route is stateless, so the provider sends `store: false` and refuses `store: true`, `store: null`, `previousResponseId`, or `conversation` — see [Errors](#errors).
 
 ## End-to-end tests
 
@@ -100,4 +218,4 @@ cp .env.example .env   # fill NEON_AI_GATEWAY_BASE_URL + NEON_AI_GATEWAY_TOKEN f
 pnpm test:e2e
 ```
 
-The matrix covers one models.dev `neon` model per family (Anthropic, OpenAI, Codex, Gemini, Meta) across `generateText`, `streamText`, `generateObject`, tool calling, and `neon.tools.imageGeneration`. It also fetches the live `/v1/models` catalog and calls every currently enabled model with both AI SDK 6 and AI SDK 7. Tests are skipped when gateway env vars are absent.
+The matrix covers one models.dev `neon` model per family (Anthropic, OpenAI, Codex, Gemini, Meta, Alibaba, Zhipu, Thinking Machines) across `generateText`, `streamText`, `generateObject`, tool calling, and `neon.tools.imageGeneration`. `generateObject` and tool calling run on the subset of families where they are verified (see [Capabilities](#capabilities)); a family whose representative id the branch does not serve is skipped rather than failed. It also fetches the live `/v1/models` catalog and calls every currently enabled model with both AI SDK 6 and AI SDK 7. Tests are skipped when gateway env vars are absent.

package/dist/index.d.ts

@@ -1,7 +1,7 @@
 import { NeonAnthropicLanguageModel } from "./lib/neon-anthropic-language-model.js";
 import { NeonChatLanguageModel } from "./lib/neon-chat-language-model.js";
-import { NEON_EXTRA_MODEL_IDS, NEON_MODELS_DEV_IDS, NeonChatModelId, NeonKnownModelId } from "./lib/neon-chat-options.js";
+import { NEON_MODELS_DEV_IDS, NeonChatModelId, NeonKnownModelId } from "./lib/neon-chat-options.js";
 import { NeonModelCapabilities, NeonModelFamily, NeonModelRoute, getNeonModelCapabilities, getNeonModelRoute } from "./lib/neon-model-capabilities.js";
 import { NeonResponsesLanguageModel } from "./lib/neon-responses-language-model.js";
 import { NeonErrorData, NeonProvider, NeonProviderSettings, createNeon, neon } from "./lib/provider.js";
-export { NEON_EXTRA_MODEL_IDS, NEON_MODELS_DEV_IDS, NeonAnthropicLanguageModel, NeonChatLanguageModel, type NeonChatModelId, type NeonErrorData, type NeonKnownModelId, type NeonModelCapabilities, type NeonModelFamily, type NeonModelRoute, type NeonProvider, type NeonProviderSettings, NeonResponsesLanguageModel, createNeon, getNeonModelCapabilities, getNeonModelRoute, neon };
+export { NEON_MODELS_DEV_IDS, NeonAnthropicLanguageModel, NeonChatLanguageModel, type NeonChatModelId, type NeonErrorData, type NeonKnownModelId, type NeonModelCapabilities, type NeonModelFamily, type NeonModelRoute, type NeonProvider, type NeonProviderSettings, NeonResponsesLanguageModel, createNeon, getNeonModelCapabilities, getNeonModelRoute, neon };

package/dist/index.js

@@ -1,7 +1,7 @@
-import { NeonAnthropicLanguageModel } from "./lib/neon-anthropic-language-model.js";
 import { getNeonModelCapabilities, getNeonModelRoute } from "./lib/neon-model-capabilities.js";
+import { NeonAnthropicLanguageModel } from "./lib/neon-anthropic-language-model.js";
 import { NeonChatLanguageModel } from "./lib/neon-chat-language-model.js";
-import { NEON_EXTRA_MODEL_IDS, NEON_MODELS_DEV_IDS } from "./lib/neon-chat-options.js";
+import { NEON_MODELS_DEV_IDS } from "./lib/neon-chat-options.js";
 import { NeonResponsesLanguageModel } from "./lib/neon-responses-language-model.js";
 import { createNeon, neon } from "./lib/provider.js";
-export { NEON_EXTRA_MODEL_IDS, NEON_MODELS_DEV_IDS, NeonAnthropicLanguageModel, NeonChatLanguageModel, NeonResponsesLanguageModel, createNeon, getNeonModelCapabilities, getNeonModelRoute, neon };
+export { NEON_MODELS_DEV_IDS, NeonAnthropicLanguageModel, NeonChatLanguageModel, NeonResponsesLanguageModel, createNeon, getNeonModelCapabilities, getNeonModelRoute, neon };

package/dist/lib/neon-anthropic-language-model.d.ts

@@ -12,6 +12,14 @@ import { LanguageModelV3, LanguageModelV3CallOptions, LanguageModelV3GenerateRes
  * (`eager_input_streaming: true`) on streaming tool calls, which the gateway
  * rejects (`Extra inputs are not permitted`). We disable it via the model's own
  * `toolStreaming` option so streaming tool calls work. Users can still override.
+ *
+ * Capability filtering runs here as well as on the unified endpoint. The
+ * upstream Anthropic model drops sampling parameters for the Claude versions it
+ * knows about, but not for ones released after it, so `claude-opus-5` and
+ * `claude-sonnet-5` reached the API and came back
+ * `` `temperature` is deprecated for this model ``. Applying our own rules keeps
+ * the behaviour tied to what the gateway was measured to accept rather than to
+ * the SDK's release date.
  */
 declare class NeonAnthropicLanguageModel extends AnthropicMessagesLanguageModel implements LanguageModelV3 {
   private withGatewayCompat;

package/dist/lib/neon-anthropic-language-model.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"neon-anthropic-language-model.d.ts","names":[],"sources":["../../src/lib/neon-anthropic-language-model.ts"],"mappings":";;;;;;;AAkBA;AACC;AAqBU;AACC;AAAR;AAKO;AACC;AAAR;AA5BK,cADI,0BAAA,SACJ,8BAAA,YACG,eADH,CAAA;EACG,QAAA,iBAAA;EAAe,UAAA,CAAA,OAAA,EAoBhB,0BApBgB,CAAA,EAqBvB,OArBuB,CAqBf,6BArBe,CAAA;oBA0BhB,6BACP,QAAQ"}
+{"version":3,"file":"neon-anthropic-language-model.d.ts","names":[],"sources":["../../src/lib/neon-anthropic-language-model.ts"],"mappings":";;;;;;;AA+BA;AACC;AA6BU;AACC;AAAR;AASO;AACC;AAAR;AAxCK;AACG;AAAe;;;;;;cAFd,0BAAA,SACJ,8BAAA,YACG;;sBA4BD,6BACP,QAAQ;oBASD,6BACP,QAAQ"}

package/dist/lib/neon-anthropic-language-model.js

@@ -1,3 +1,4 @@
+import { applyNeonCapabilities, mergeStreamStartWarnings } from "./neon-capabilities.js";
 import { AnthropicMessagesLanguageModel } from "@ai-sdk/anthropic/internal";
 //#region src/lib/neon-anthropic-language-model.ts
 /**
@@ -9,27 +10,52 @@ import { AnthropicMessagesLanguageModel } from "@ai-sdk/anthropic/internal";
 * (`eager_input_streaming: true`) on streaming tool calls, which the gateway
 * rejects (`Extra inputs are not permitted`). We disable it via the model's own
 * `toolStreaming` option so streaming tool calls work. Users can still override.
+*
+* Capability filtering runs here as well as on the unified endpoint. The
+* upstream Anthropic model drops sampling parameters for the Claude versions it
+* knows about, but not for ones released after it, so `claude-opus-5` and
+* `claude-sonnet-5` reached the API and came back
+* `` `temperature` is deprecated for this model ``. Applying our own rules keeps
+* the behaviour tied to what the gateway was measured to accept rather than to
+* the SDK's release date.
 */
 var NeonAnthropicLanguageModel = class extends AnthropicMessagesLanguageModel {
 	withGatewayCompat(options) {
-		const anthropic = options.providerOptions?.anthropic;
-		if (anthropic != null && "toolStreaming" in anthropic) return options;
+		const { options: adjusted, warnings } = applyNeonCapabilities(this.modelId, options);
+		const anthropic = adjusted.providerOptions?.anthropic;
+		if (anthropic != null && "toolStreaming" in anthropic) return {
+			options: adjusted,
+			warnings
+		};
 		return {
-			...options,
-			providerOptions: {
-				...options.providerOptions,
-				anthropic: {
-					...anthropic,
-					toolStreaming: false
+			options: {
+				...adjusted,
+				providerOptions: {
+					...adjusted.providerOptions,
+					anthropic: {
+						...anthropic,
+						toolStreaming: false
+					}
 				}
-			}
+			},
+			warnings
 		};
 	}
-	doGenerate(options) {
-		return super.doGenerate(this.withGatewayCompat(options));
+	async doGenerate(options) {
+		const { options: adjusted, warnings } = this.withGatewayCompat(options);
+		const result = await super.doGenerate(adjusted);
+		return warnings.length > 0 ? {
+			...result,
+			warnings: [...warnings, ...result.warnings]
+		} : result;
 	}
-	doStream(options) {
-		return super.doStream(this.withGatewayCompat(options));
+	async doStream(options) {
+		const { options: adjusted, warnings } = this.withGatewayCompat(options);
+		const result = await super.doStream(adjusted);
+		return warnings.length > 0 ? {
+			...result,
+			stream: result.stream.pipeThrough(mergeStreamStartWarnings(warnings))
+		} : result;
 	}
 };
 //#endregion

package/dist/lib/neon-anthropic-language-model.js.map

@@ -1 +1 @@
-{"version":3,"file":"neon-anthropic-language-model.js","names":[],"sources":["../../src/lib/neon-anthropic-language-model.ts"],"sourcesContent":["import { AnthropicMessagesLanguageModel } from \"@ai-sdk/anthropic/internal\";\nimport type {\n\tLanguageModelV3,\n\tLanguageModelV3CallOptions,\n\tLanguageModelV3GenerateResult,\n\tLanguageModelV3StreamResult,\n} from \"@ai-sdk/provider\";\n\n/**\n * Language model for Anthropic models served via the Neon AI Gateway's native\n * Messages API (`/anthropic/v1/messages`). Unlocks streaming\n * structured output and native reasoning for Claude.\n *\n * The shared Anthropic model defaults to fine-grained tool-input streaming\n * (`eager_input_streaming: true`) on streaming tool calls, which the gateway\n * rejects (`Extra inputs are not permitted`). We disable it via the model's own\n * `toolStreaming` option so streaming tool calls work. Users can still override.\n */\nexport class NeonAnthropicLanguageModel\n\textends AnthropicMessagesLanguageModel\n\timplements LanguageModelV3\n{\n\tprivate withGatewayCompat(\n\t\toptions: LanguageModelV3CallOptions,\n\t): LanguageModelV3CallOptions {\n\t\tconst anthropic = options.providerOptions?.anthropic;\n\t\t// Respect an explicit user setting.\n\t\tif (anthropic != null && \"toolStreaming\" in anthropic) {\n\t\t\treturn options;\n\t\t}\n\t\treturn {\n\t\t\t...options,\n\t\t\tproviderOptions: {\n\t\t\t\t...options.providerOptions,\n\t\t\t\tanthropic: { ...anthropic, toolStreaming: false },\n\t\t\t},\n\t\t};\n\t}\n\n\toverride doGenerate(\n\t\toptions: LanguageModelV3CallOptions,\n\t): Promise<LanguageModelV3GenerateResult> {\n\t\treturn super.doGenerate(this.withGatewayCompat(options));\n\t}\n\n\toverride doStream(\n\t\toptions: LanguageModelV3CallOptions,\n\t): Promise<LanguageModelV3StreamResult> {\n\t\treturn super.doStream(this.withGatewayCompat(options));\n\t}\n}\n"],"mappings":";;;;;;;;;;;;AAkBA,IAAa,6BAAb,cACS,+BAET;CACC,kBACC,SAC6B;EAC7B,MAAM,YAAY,QAAQ,iBAAiB;EAE3C,IAAI,aAAa,QAAQ,mBAAmB,WAC3C,OAAO;EAER,OAAO;GACN,GAAG;GACH,iBAAiB;IAChB,GAAG,QAAQ;IACX,WAAW;KAAE,GAAG;KAAW,eAAe;IAAM;GACjD;EACD;CACD;CAEA,WACC,SACyC;EACzC,OAAO,MAAM,WAAW,KAAK,kBAAkB,OAAO,CAAC;CACxD;CAEA,SACC,SACuC;EACvC,OAAO,MAAM,SAAS,KAAK,kBAAkB,OAAO,CAAC;CACtD;AACD"}
+{"version":3,"file":"neon-anthropic-language-model.js","names":[],"sources":["../../src/lib/neon-anthropic-language-model.ts"],"sourcesContent":["import { AnthropicMessagesLanguageModel } from \"@ai-sdk/anthropic/internal\";\nimport type {\n\tLanguageModelV3,\n\tLanguageModelV3CallOptions,\n\tLanguageModelV3GenerateResult,\n\tLanguageModelV3StreamResult,\n\tSharedV3Warning,\n} from \"@ai-sdk/provider\";\nimport {\n\tapplyNeonCapabilities,\n\tmergeStreamStartWarnings,\n} from \"./neon-capabilities.js\";\n\n/**\n * Language model for Anthropic models served via the Neon AI Gateway's native\n * Messages API (`/anthropic/v1/messages`). Unlocks streaming\n * structured output and native reasoning for Claude.\n *\n * The shared Anthropic model defaults to fine-grained tool-input streaming\n * (`eager_input_streaming: true`) on streaming tool calls, which the gateway\n * rejects (`Extra inputs are not permitted`). We disable it via the model's own\n * `toolStreaming` option so streaming tool calls work. Users can still override.\n *\n * Capability filtering runs here as well as on the unified endpoint. The\n * upstream Anthropic model drops sampling parameters for the Claude versions it\n * knows about, but not for ones released after it, so `claude-opus-5` and\n * `claude-sonnet-5` reached the API and came back\n * `` `temperature` is deprecated for this model ``. Applying our own rules keeps\n * the behaviour tied to what the gateway was measured to accept rather than to\n * the SDK's release date.\n */\nexport class NeonAnthropicLanguageModel\n\textends AnthropicMessagesLanguageModel\n\timplements LanguageModelV3\n{\n\tprivate withGatewayCompat(options: LanguageModelV3CallOptions): {\n\t\toptions: LanguageModelV3CallOptions;\n\t\twarnings: SharedV3Warning[];\n\t} {\n\t\tconst { options: adjusted, warnings } = applyNeonCapabilities(\n\t\t\tthis.modelId,\n\t\t\toptions,\n\t\t);\n\t\tconst anthropic = adjusted.providerOptions?.anthropic;\n\t\t// Respect an explicit user setting.\n\t\tif (anthropic != null && \"toolStreaming\" in anthropic) {\n\t\t\treturn { options: adjusted, warnings };\n\t\t}\n\t\treturn {\n\t\t\toptions: {\n\t\t\t\t...adjusted,\n\t\t\t\tproviderOptions: {\n\t\t\t\t\t...adjusted.providerOptions,\n\t\t\t\t\tanthropic: { ...anthropic, toolStreaming: false },\n\t\t\t\t},\n\t\t\t},\n\t\t\twarnings,\n\t\t};\n\t}\n\n\toverride async doGenerate(\n\t\toptions: LanguageModelV3CallOptions,\n\t): Promise<LanguageModelV3GenerateResult> {\n\t\tconst { options: adjusted, warnings } = this.withGatewayCompat(options);\n\t\tconst result = await super.doGenerate(adjusted);\n\t\treturn warnings.length > 0\n\t\t\t? { ...result, warnings: [...warnings, ...result.warnings] }\n\t\t\t: result;\n\t}\n\n\toverride async doStream(\n\t\toptions: LanguageModelV3CallOptions,\n\t): Promise<LanguageModelV3StreamResult> {\n\t\tconst { options: adjusted, warnings } = this.withGatewayCompat(options);\n\t\tconst result = await super.doStream(adjusted);\n\t\treturn warnings.length > 0\n\t\t\t? {\n\t\t\t\t\t...result,\n\t\t\t\t\tstream: result.stream.pipeThrough(\n\t\t\t\t\t\tmergeStreamStartWarnings(warnings),\n\t\t\t\t\t),\n\t\t\t\t}\n\t\t\t: result;\n\t}\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;AA+BA,IAAa,6BAAb,cACS,+BAET;CACC,kBAA0B,SAGxB;EACD,MAAM,EAAE,SAAS,UAAU,aAAa,sBACvC,KAAK,SACL,OACD;EACA,MAAM,YAAY,SAAS,iBAAiB;EAE5C,IAAI,aAAa,QAAQ,mBAAmB,WAC3C,OAAO;GAAE,SAAS;GAAU;EAAS;EAEtC,OAAO;GACN,SAAS;IACR,GAAG;IACH,iBAAiB;KAChB,GAAG,SAAS;KACZ,WAAW;MAAE,GAAG;MAAW,eAAe;KAAM;IACjD;GACD;GACA;EACD;CACD;CAEA,MAAe,WACd,SACyC;EACzC,MAAM,EAAE,SAAS,UAAU,aAAa,KAAK,kBAAkB,OAAO;EACtE,MAAM,SAAS,MAAM,MAAM,WAAW,QAAQ;EAC9C,OAAO,SAAS,SAAS,IACtB;GAAE,GAAG;GAAQ,UAAU,CAAC,GAAG,UAAU,GAAG,OAAO,QAAQ;EAAE,IACzD;CACJ;CAEA,MAAe,SACd,SACuC;EACvC,MAAM,EAAE,SAAS,UAAU,aAAa,KAAK,kBAAkB,OAAO;EACtE,MAAM,SAAS,MAAM,MAAM,SAAS,QAAQ;EAC5C,OAAO,SAAS,SAAS,IACtB;GACA,GAAG;GACH,QAAQ,OAAO,OAAO,YACrB,yBAAyB,QAAQ,CAClC;EACD,IACC;CACJ;AACD"}

package/dist/lib/neon-capabilities.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"neon-capabilities.d.ts","names":[],"sources":["../../src/lib/neon-capabilities.ts"],"mappings":";;;;;;AAaA;AAAqC;AAE3B;AACI,iBAHE,qBAAA,CAGF,OAAA,EAAA,MAAA,EAAA,OAAA,EADJ,0BACI,CAAA,EAAA;EAAsC,OAAA,EAAtC,0BAAsC;EAAe,QAAA,EAAf,eAAe,EAAA;AA4FnE,CAAA;AAAwC;AAAQ;AAAiB;AAAA,iBAAjD,wBAAA,CAAiD,KAAA,EAAjB,eAAiB,EAAA,CAAA,EAAA,eAAA,CAAA,yBAAA,EAAA,yBAAA,CAAA"}
+{"version":3,"file":"neon-capabilities.d.ts","names":[],"sources":["../../src/lib/neon-capabilities.ts"],"mappings":";;;;;;AAoBA;AAAqC;AAE3B;AACI,iBAHE,qBAAA,CAGF,OAAA,EAAA,MAAA,EAAA,OAAA,EADJ,0BACI,CAAA,EAAA;EAAsC,OAAA,EAAtC,0BAAsC;EAAe,QAAA,EAAf,eAAe,EAAA;AA8GnE,CAAA;AAAwC;AAAQ;AAAiB;AAAA,iBAAjD,wBAAA,CAAiD,KAAA,EAAjB,eAAiB,EAAA,CAAA,EAAA,eAAA,CAAA,yBAAA,EAAA,yBAAA,CAAA"}

package/dist/lib/neon-capabilities.js

@@ -1,5 +1,8 @@
 import { getNeonModelCapabilities } from "./neon-model-capabilities.js";
 //#region src/lib/neon-capabilities.ts
+const PENALTY_DETAILS = "Only Gemini accepts penalties on the gateway's unified endpoint, so it was dropped rather than sent.";
+const GENERIC_DETAILS = "The Neon AI Gateway rejects this parameter for this model, so it was dropped rather than sent.";
+const REASONING_EFFORT_DETAILS = "This model does not take the OpenAI reasoning_effort field, so it was dropped rather than sent.";
 /**
 * Drop call options the resolved model's upstream backend does not accept and
 * collect a warning for each, so callers get a clear signal instead of a hard
@@ -9,20 +12,23 @@ function applyNeonCapabilities(modelId, options) {
 	const caps = getNeonModelCapabilities(modelId);
 	const warnings = [];
 	const patch = {};
-	const dropUnsupported = (feature) => {
+	const drop = (feature, details, type = "unsupported") => {
 		warnings.push({
-			type: "unsupported",
+			type,
 			feature,
-			details: `${feature} is not supported by the Neon AI Gateway for ${caps.family} model "${modelId}" and was dropped.`
+			details
 		});
 	};
+	const unrecognizedClaude = caps.claudeSamplingUnrecognized === true;
+	const samplingDetails = unrecognizedClaude ? "This Claude version was not recognised. Claude 4.7 and newer reject sampling parameters, so it was dropped as a precaution; set `providerOptions.anthropic.effort` instead." : caps.family === "anthropic" ? "Claude 4.7 and newer reject sampling parameters, so it was dropped rather than sent. Use `providerOptions.anthropic.effort` (low | medium | high | xhigh | max) to steer these models." : "The Neon AI Gateway rejects this parameter for this model, so it was dropped rather than sent.";
+	const samplingType = unrecognizedClaude ? "compatibility" : "unsupported";
 	if (options.temperature != null && !caps.supportsTemperature) {
 		patch.temperature = void 0;
-		dropUnsupported("temperature");
+		drop("temperature", samplingDetails, samplingType);
 	}
 	if (options.topP != null && !caps.supportsTopP) {
 		patch.topP = void 0;
-		dropUnsupported("topP");
+		drop("topP", samplingDetails, samplingType);
 	}
 	const effectiveTemperature = "temperature" in patch ? patch.temperature : options.temperature;
 	const effectiveTopP = "topP" in patch ? patch.topP : options.topP;
@@ -31,29 +37,29 @@ function applyNeonCapabilities(modelId, options) {
 		warnings.push({
 			type: "compatibility",
 			feature: "topP",
-			details: `${caps.family} models accept only one of temperature or topP; topP was dropped.`
+			details: "Anthropic models accept only one of temperature or topP; topP was dropped and temperature kept."
 		});
 	}
 	if (options.frequencyPenalty != null && !caps.supportsPenalties) {
 		patch.frequencyPenalty = void 0;
-		dropUnsupported("frequencyPenalty");
+		drop("frequencyPenalty", PENALTY_DETAILS);
 	}
 	if (options.presencePenalty != null && !caps.supportsPenalties) {
 		patch.presencePenalty = void 0;
-		dropUnsupported("presencePenalty");
+		drop("presencePenalty", PENALTY_DETAILS);
 	}
 	if (options.seed != null && !caps.supportsSeed) {
 		patch.seed = void 0;
-		dropUnsupported("seed");
+		drop("seed", GENERIC_DETAILS);
 	}
 	if (options.stopSequences != null && !caps.supportsStopSequences) {
 		patch.stopSequences = void 0;
-		dropUnsupported("stopSequences");
+		drop("stopSequences", GENERIC_DETAILS);
 	}
 	if (!caps.supportsReasoningEffort) {
 		const { providerOptions } = options;
 		if (providerOptions != null && Object.values(providerOptions).some((group) => "reasoningEffort" in group && group.reasoningEffort != null)) {
-			dropUnsupported("reasoningEffort");
+			drop("reasoningEffort", REASONING_EFFORT_DETAILS);
 			if (providerOptions != null) {
 				const cleaned = {};
 				for (const [key, group] of Object.entries(providerOptions)) if ("reasoningEffort" in group) {

package/dist/lib/neon-capabilities.js.map

@@ -1 +1 @@
-{"version":3,"file":"neon-capabilities.js","names":[],"sources":["../../src/lib/neon-capabilities.ts"],"sourcesContent":["import type {\n\tLanguageModelV3CallOptions,\n\tLanguageModelV3StreamPart,\n\tSharedV3ProviderOptions,\n\tSharedV3Warning,\n} from \"@ai-sdk/provider\";\nimport { getNeonModelCapabilities } from \"./neon-model-capabilities.js\";\n\n/**\n * Drop call options the resolved model's upstream backend does not accept and\n * collect a warning for each, so callers get a clear signal instead of a hard\n * `400` from the gateway.\n */\nexport function applyNeonCapabilities(\n\tmodelId: string,\n\toptions: LanguageModelV3CallOptions,\n): { options: LanguageModelV3CallOptions; warnings: SharedV3Warning[] } {\n\tconst caps = getNeonModelCapabilities(modelId);\n\tconst warnings: SharedV3Warning[] = [];\n\tconst patch: Partial<LanguageModelV3CallOptions> = {};\n\n\tconst dropUnsupported = (feature: string) => {\n\t\twarnings.push({\n\t\t\ttype: \"unsupported\",\n\t\t\tfeature,\n\t\t\tdetails: `${feature} is not supported by the Neon AI Gateway for ${caps.family} model \"${modelId}\" and was dropped.`,\n\t\t});\n\t};\n\n\tif (options.temperature != null && !caps.supportsTemperature) {\n\t\tpatch.temperature = undefined;\n\t\tdropUnsupported(\"temperature\");\n\t}\n\tif (options.topP != null && !caps.supportsTopP) {\n\t\tpatch.topP = undefined;\n\t\tdropUnsupported(\"topP\");\n\t}\n\n\t// Anthropic-style models accept only one of temperature / topP.\n\tconst effectiveTemperature =\n\t\t\"temperature\" in patch ? patch.temperature : options.temperature;\n\tconst effectiveTopP = \"topP\" in patch ? patch.topP : options.topP;\n\tif (\n\t\tcaps.temperatureTopPMutuallyExclusive &&\n\t\teffectiveTemperature != null &&\n\t\teffectiveTopP != null\n\t) {\n\t\tpatch.topP = undefined;\n\t\twarnings.push({\n\t\t\ttype: \"compatibility\",\n\t\t\tfeature: \"topP\",\n\t\t\tdetails: `${caps.family} models accept only one of temperature or topP; topP was dropped.`,\n\t\t});\n\t}\n\n\tif (options.frequencyPenalty != null && !caps.supportsPenalties) {\n\t\tpatch.frequencyPenalty = undefined;\n\t\tdropUnsupported(\"frequencyPenalty\");\n\t}\n\tif (options.presencePenalty != null && !caps.supportsPenalties) {\n\t\tpatch.presencePenalty = undefined;\n\t\tdropUnsupported(\"presencePenalty\");\n\t}\n\tif (options.seed != null && !caps.supportsSeed) {\n\t\tpatch.seed = undefined;\n\t\tdropUnsupported(\"seed\");\n\t}\n\n\tif (options.stopSequences != null && !caps.supportsStopSequences) {\n\t\tpatch.stopSequences = undefined;\n\t\tdropUnsupported(\"stopSequences\");\n\t}\n\n\tif (!caps.supportsReasoningEffort) {\n\t\tconst { providerOptions } = options;\n\t\tconst hasProviderEffort =\n\t\t\tproviderOptions != null &&\n\t\t\tObject.values(providerOptions).some(\n\t\t\t\t(group) =>\n\t\t\t\t\t\"reasoningEffort\" in group && group.reasoningEffort != null,\n\t\t\t);\n\n\t\tif (hasProviderEffort) {\n\t\t\tdropUnsupported(\"reasoningEffort\");\n\t\t\tif (providerOptions != null) {\n\t\t\t\tconst cleaned: SharedV3ProviderOptions = {};\n\t\t\t\tfor (const [key, group] of Object.entries(providerOptions)) {\n\t\t\t\t\tif (\"reasoningEffort\" in group) {\n\t\t\t\t\t\tconst { reasoningEffort: _removed, ...rest } = group;\n\t\t\t\t\t\tcleaned[key] = rest;\n\t\t\t\t\t} else {\n\t\t\t\t\t\tcleaned[key] = group;\n\t\t\t\t\t}\n\t\t\t\t}\n\t\t\t\tpatch.providerOptions = cleaned;\n\t\t\t}\n\t\t}\n\t}\n\n\tif (Object.keys(patch).length === 0) {\n\t\treturn { options, warnings };\n\t}\n\treturn { options: { ...options, ...patch }, warnings };\n}\n\n/**\n * Merge additional warnings into the `stream-start` part of a model stream.\n */\nexport function mergeStreamStartWarnings(extra: SharedV3Warning[]) {\n\tlet merged = false;\n\treturn new TransformStream<\n\t\tLanguageModelV3StreamPart,\n\t\tLanguageModelV3StreamPart\n\t>({\n\t\ttransform(part, controller) {\n\t\t\tif (!merged && part.type === \"stream-start\") {\n\t\t\t\tmerged = true;\n\t\t\t\tcontroller.enqueue({\n\t\t\t\t\t...part,\n\t\t\t\t\twarnings: [...extra, ...part.warnings],\n\t\t\t\t});\n\t\t\t} else {\n\t\t\t\tcontroller.enqueue(part);\n\t\t\t}\n\t\t},\n\t});\n}\n"],"mappings":";;;;;;;AAaA,SAAgB,sBACf,SACA,SACuE;CACvE,MAAM,OAAO,yBAAyB,OAAO;CAC7C,MAAM,WAA8B,CAAC;CACrC,MAAM,QAA6C,CAAC;CAEpD,MAAM,mBAAmB,YAAoB;EAC5C,SAAS,KAAK;GACb,MAAM;GACN;GACA,SAAS,GAAG,QAAQ,+CAA+C,KAAK,OAAO,UAAU,QAAQ;EAClG,CAAC;CACF;CAEA,IAAI,QAAQ,eAAe,QAAQ,CAAC,KAAK,qBAAqB;EAC7D,MAAM,cAAc,KAAA;EACpB,gBAAgB,aAAa;CAC9B;CACA,IAAI,QAAQ,QAAQ,QAAQ,CAAC,KAAK,cAAc;EAC/C,MAAM,OAAO,KAAA;EACb,gBAAgB,MAAM;CACvB;CAGA,MAAM,uBACL,iBAAiB,QAAQ,MAAM,cAAc,QAAQ;CACtD,MAAM,gBAAgB,UAAU,QAAQ,MAAM,OAAO,QAAQ;CAC7D,IACC,KAAK,oCACL,wBAAwB,QACxB,iBAAiB,MAChB;EACD,MAAM,OAAO,KAAA;EACb,SAAS,KAAK;GACb,MAAM;GACN,SAAS;GACT,SAAS,GAAG,KAAK,OAAO;EACzB,CAAC;CACF;CAEA,IAAI,QAAQ,oBAAoB,QAAQ,CAAC,KAAK,mBAAmB;EAChE,MAAM,mBAAmB,KAAA;EACzB,gBAAgB,kBAAkB;CACnC;CACA,IAAI,QAAQ,mBAAmB,QAAQ,CAAC,KAAK,mBAAmB;EAC/D,MAAM,kBAAkB,KAAA;EACxB,gBAAgB,iBAAiB;CAClC;CACA,IAAI,QAAQ,QAAQ,QAAQ,CAAC,KAAK,cAAc;EAC/C,MAAM,OAAO,KAAA;EACb,gBAAgB,MAAM;CACvB;CAEA,IAAI,QAAQ,iBAAiB,QAAQ,CAAC,KAAK,uBAAuB;EACjE,MAAM,gBAAgB,KAAA;EACtB,gBAAgB,eAAe;CAChC;CAEA,IAAI,CAAC,KAAK,yBAAyB;EAClC,MAAM,EAAE,oBAAoB;EAQ5B,IANC,mBAAmB,QACnB,OAAO,OAAO,eAAe,CAAC,CAAC,MAC7B,UACA,qBAAqB,SAAS,MAAM,mBAAmB,IACzD,GAEsB;GACtB,gBAAgB,iBAAiB;GACjC,IAAI,mBAAmB,MAAM;IAC5B,MAAM,UAAmC,CAAC;IAC1C,KAAK,MAAM,CAAC,KAAK,UAAU,OAAO,QAAQ,eAAe,GACxD,IAAI,qBAAqB,OAAO;KAC/B,MAAM,EAAE,iBAAiB,UAAU,GAAG,SAAS;KAC/C,QAAQ,OAAO;IAChB,OACC,QAAQ,OAAO;IAGjB,MAAM,kBAAkB;GACzB;EACD;CACD;CAEA,IAAI,OAAO,KAAK,KAAK,CAAC,CAAC,WAAW,GACjC,OAAO;EAAE;EAAS;CAAS;CAE5B,OAAO;EAAE,SAAS;GAAE,GAAG;GAAS,GAAG;EAAM;EAAG;CAAS;AACtD;;;;AAKA,SAAgB,yBAAyB,OAA0B;CAClE,IAAI,SAAS;CACb,OAAO,IAAI,gBAGT,EACD,UAAU,MAAM,YAAY;EAC3B,IAAI,CAAC,UAAU,KAAK,SAAS,gBAAgB;GAC5C,SAAS;GACT,WAAW,QAAQ;IAClB,GAAG;IACH,UAAU,CAAC,GAAG,OAAO,GAAG,KAAK,QAAQ;GACtC,CAAC;EACF,OACC,WAAW,QAAQ,IAAI;CAEzB,EACD,CAAC;AACF"}
+{"version":3,"file":"neon-capabilities.js","names":[],"sources":["../../src/lib/neon-capabilities.ts"],"sourcesContent":["import type {\n\tLanguageModelV3CallOptions,\n\tLanguageModelV3StreamPart,\n\tSharedV3ProviderOptions,\n\tSharedV3Warning,\n} from \"@ai-sdk/provider\";\nimport { getNeonModelCapabilities } from \"./neon-model-capabilities.js\";\n\nconst PENALTY_DETAILS =\n\t\"Only Gemini accepts penalties on the gateway's unified endpoint, so it was dropped rather than sent.\";\nconst GENERIC_DETAILS =\n\t\"The Neon AI Gateway rejects this parameter for this model, so it was dropped rather than sent.\";\nconst REASONING_EFFORT_DETAILS =\n\t\"This model does not take the OpenAI reasoning_effort field, so it was dropped rather than sent.\";\n\n/**\n * Drop call options the resolved model's upstream backend does not accept and\n * collect a warning for each, so callers get a clear signal instead of a hard\n * `400` from the gateway.\n */\nexport function applyNeonCapabilities(\n\tmodelId: string,\n\toptions: LanguageModelV3CallOptions,\n): { options: LanguageModelV3CallOptions; warnings: SharedV3Warning[] } {\n\tconst caps = getNeonModelCapabilities(modelId);\n\tconst warnings: SharedV3Warning[] = [];\n\tconst patch: Partial<LanguageModelV3CallOptions> = {};\n\n\t// The AI SDK prints the provider, the model id and the feature name ahead of\n\t// `details`, so each reason says only what those cannot: why the gateway\n\t// refuses it, and what to use instead. Each carries its own \"was dropped\"\n\t// clause rather than a shared suffix — appending one everywhere restated the\n\t// penalty reason and asserted a 400 for the unrecognised case, which is the\n\t// overstatement this wording exists to avoid.\n\tconst drop = (\n\t\tfeature: string,\n\t\tdetails: string,\n\t\ttype: \"unsupported\" | \"compatibility\" = \"unsupported\",\n\t) => {\n\t\twarnings.push({ type, feature, details });\n\t};\n\n\t// Precautionary rather than measured, so it is flagged as a compatibility\n\t// choice: the SDK renders `unsupported` as a flat \"is not supported\", which\n\t// would contradict the hedge in the sentence that follows it.\n\tconst unrecognizedClaude = caps.claudeSamplingUnrecognized === true;\n\tconst samplingDetails = unrecognizedClaude\n\t\t? \"This Claude version was not recognised. Claude 4.7 and newer reject sampling parameters, so it was dropped as a precaution; set `providerOptions.anthropic.effort` instead.\"\n\t\t: caps.family === \"anthropic\"\n\t\t\t? \"Claude 4.7 and newer reject sampling parameters, so it was dropped rather than sent. Use `providerOptions.anthropic.effort` (low | medium | high | xhigh | max) to steer these models.\"\n\t\t\t: \"The Neon AI Gateway rejects this parameter for this model, so it was dropped rather than sent.\";\n\tconst samplingType = unrecognizedClaude ? \"compatibility\" : \"unsupported\";\n\n\tif (options.temperature != null && !caps.supportsTemperature) {\n\t\tpatch.temperature = undefined;\n\t\tdrop(\"temperature\", samplingDetails, samplingType);\n\t}\n\tif (options.topP != null && !caps.supportsTopP) {\n\t\tpatch.topP = undefined;\n\t\tdrop(\"topP\", samplingDetails, samplingType);\n\t}\n\n\t// Anthropic-style models accept only one of temperature / topP.\n\tconst effectiveTemperature =\n\t\t\"temperature\" in patch ? patch.temperature : options.temperature;\n\tconst effectiveTopP = \"topP\" in patch ? patch.topP : options.topP;\n\tif (\n\t\tcaps.temperatureTopPMutuallyExclusive &&\n\t\teffectiveTemperature != null &&\n\t\teffectiveTopP != null\n\t) {\n\t\tpatch.topP = undefined;\n\t\twarnings.push({\n\t\t\ttype: \"compatibility\",\n\t\t\tfeature: \"topP\",\n\t\t\tdetails:\n\t\t\t\t\"Anthropic models accept only one of temperature or topP; topP was dropped and temperature kept.\",\n\t\t});\n\t}\n\n\tif (options.frequencyPenalty != null && !caps.supportsPenalties) {\n\t\tpatch.frequencyPenalty = undefined;\n\t\tdrop(\"frequencyPenalty\", PENALTY_DETAILS);\n\t}\n\tif (options.presencePenalty != null && !caps.supportsPenalties) {\n\t\tpatch.presencePenalty = undefined;\n\t\tdrop(\"presencePenalty\", PENALTY_DETAILS);\n\t}\n\tif (options.seed != null && !caps.supportsSeed) {\n\t\tpatch.seed = undefined;\n\t\tdrop(\"seed\", GENERIC_DETAILS);\n\t}\n\n\tif (options.stopSequences != null && !caps.supportsStopSequences) {\n\t\tpatch.stopSequences = undefined;\n\t\tdrop(\"stopSequences\", GENERIC_DETAILS);\n\t}\n\n\tif (!caps.supportsReasoningEffort) {\n\t\tconst { providerOptions } = options;\n\t\tconst hasProviderEffort =\n\t\t\tproviderOptions != null &&\n\t\t\tObject.values(providerOptions).some(\n\t\t\t\t(group) =>\n\t\t\t\t\t\"reasoningEffort\" in group && group.reasoningEffort != null,\n\t\t\t);\n\n\t\tif (hasProviderEffort) {\n\t\t\tdrop(\"reasoningEffort\", REASONING_EFFORT_DETAILS);\n\t\t\tif (providerOptions != null) {\n\t\t\t\tconst cleaned: SharedV3ProviderOptions = {};\n\t\t\t\tfor (const [key, group] of Object.entries(providerOptions)) {\n\t\t\t\t\tif (\"reasoningEffort\" in group) {\n\t\t\t\t\t\tconst { reasoningEffort: _removed, ...rest } = group;\n\t\t\t\t\t\tcleaned[key] = rest;\n\t\t\t\t\t} else {\n\t\t\t\t\t\tcleaned[key] = group;\n\t\t\t\t\t}\n\t\t\t\t}\n\t\t\t\tpatch.providerOptions = cleaned;\n\t\t\t}\n\t\t}\n\t}\n\n\tif (Object.keys(patch).length === 0) {\n\t\treturn { options, warnings };\n\t}\n\treturn { options: { ...options, ...patch }, warnings };\n}\n\n/**\n * Merge additional warnings into the `stream-start` part of a model stream.\n */\nexport function mergeStreamStartWarnings(extra: SharedV3Warning[]) {\n\tlet merged = false;\n\treturn new TransformStream<\n\t\tLanguageModelV3StreamPart,\n\t\tLanguageModelV3StreamPart\n\t>({\n\t\ttransform(part, controller) {\n\t\t\tif (!merged && part.type === \"stream-start\") {\n\t\t\t\tmerged = true;\n\t\t\t\tcontroller.enqueue({\n\t\t\t\t\t...part,\n\t\t\t\t\twarnings: [...extra, ...part.warnings],\n\t\t\t\t});\n\t\t\t} else {\n\t\t\t\tcontroller.enqueue(part);\n\t\t\t}\n\t\t},\n\t});\n}\n"],"mappings":";;AAQA,MAAM,kBACL;AACD,MAAM,kBACL;AACD,MAAM,2BACL;;;;;;AAOD,SAAgB,sBACf,SACA,SACuE;CACvE,MAAM,OAAO,yBAAyB,OAAO;CAC7C,MAAM,WAA8B,CAAC;CACrC,MAAM,QAA6C,CAAC;CAQpD,MAAM,QACL,SACA,SACA,OAAwC,kBACpC;EACJ,SAAS,KAAK;GAAE;GAAM;GAAS;EAAQ,CAAC;CACzC;CAKA,MAAM,qBAAqB,KAAK,+BAA+B;CAC/D,MAAM,kBAAkB,qBACrB,gLACA,KAAK,WAAW,cACf,2LACA;CACJ,MAAM,eAAe,qBAAqB,kBAAkB;CAE5D,IAAI,QAAQ,eAAe,QAAQ,CAAC,KAAK,qBAAqB;EAC7D,MAAM,cAAc,KAAA;EACpB,KAAK,eAAe,iBAAiB,YAAY;CAClD;CACA,IAAI,QAAQ,QAAQ,QAAQ,CAAC,KAAK,cAAc;EAC/C,MAAM,OAAO,KAAA;EACb,KAAK,QAAQ,iBAAiB,YAAY;CAC3C;CAGA,MAAM,uBACL,iBAAiB,QAAQ,MAAM,cAAc,QAAQ;CACtD,MAAM,gBAAgB,UAAU,QAAQ,MAAM,OAAO,QAAQ;CAC7D,IACC,KAAK,oCACL,wBAAwB,QACxB,iBAAiB,MAChB;EACD,MAAM,OAAO,KAAA;EACb,SAAS,KAAK;GACb,MAAM;GACN,SAAS;GACT,SACC;EACF,CAAC;CACF;CAEA,IAAI,QAAQ,oBAAoB,QAAQ,CAAC,KAAK,mBAAmB;EAChE,MAAM,mBAAmB,KAAA;EACzB,KAAK,oBAAoB,eAAe;CACzC;CACA,IAAI,QAAQ,mBAAmB,QAAQ,CAAC,KAAK,mBAAmB;EAC/D,MAAM,kBAAkB,KAAA;EACxB,KAAK,mBAAmB,eAAe;CACxC;CACA,IAAI,QAAQ,QAAQ,QAAQ,CAAC,KAAK,cAAc;EAC/C,MAAM,OAAO,KAAA;EACb,KAAK,QAAQ,eAAe;CAC7B;CAEA,IAAI,QAAQ,iBAAiB,QAAQ,CAAC,KAAK,uBAAuB;EACjE,MAAM,gBAAgB,KAAA;EACtB,KAAK,iBAAiB,eAAe;CACtC;CAEA,IAAI,CAAC,KAAK,yBAAyB;EAClC,MAAM,EAAE,oBAAoB;EAQ5B,IANC,mBAAmB,QACnB,OAAO,OAAO,eAAe,CAAC,CAAC,MAC7B,UACA,qBAAqB,SAAS,MAAM,mBAAmB,IACzD,GAEsB;GACtB,KAAK,mBAAmB,wBAAwB;GAChD,IAAI,mBAAmB,MAAM;IAC5B,MAAM,UAAmC,CAAC;IAC1C,KAAK,MAAM,CAAC,KAAK,UAAU,OAAO,QAAQ,eAAe,GACxD,IAAI,qBAAqB,OAAO;KAC/B,MAAM,EAAE,iBAAiB,UAAU,GAAG,SAAS;KAC/C,QAAQ,OAAO;IAChB,OACC,QAAQ,OAAO;IAGjB,MAAM,kBAAkB;GACzB;EACD;CACD;CAEA,IAAI,OAAO,KAAK,KAAK,CAAC,CAAC,WAAW,GACjC,OAAO;EAAE;EAAS;CAAS;CAE5B,OAAO;EAAE,SAAS;GAAE,GAAG;GAAS,GAAG;EAAM;EAAG;CAAS;AACtD;;;;AAKA,SAAgB,yBAAyB,OAA0B;CAClE,IAAI,SAAS;CACb,OAAO,IAAI,gBAGT,EACD,UAAU,MAAM,YAAY;EAC3B,IAAI,CAAC,UAAU,KAAK,SAAS,gBAAgB;GAC5C,SAAS;GACT,WAAW,QAAQ;IAClB,GAAG;IACH,UAAU,CAAC,GAAG,OAAO,GAAG,KAAK,QAAQ;GACtC,CAAC;EACF,OACC,WAAW,QAAQ,IAAI;CAEzB,EACD,CAAC;AACF"}

package/dist/lib/neon-chat-options.d.ts

@@ -1,15 +1,8 @@
 //#region src/lib/neon-chat-options.d.ts
 /** The models.dev `neon` catalog (canonical, unprefixed ids). */
-declare const NEON_MODELS_DEV_IDS: readonly ["claude-opus-4-7", "claude-opus-4-6", "claude-opus-4-5", "claude-opus-4-1", "claude-sonnet-4-6", "claude-sonnet-4-5", "claude-sonnet-4", "claude-haiku-4-5", "gpt-5", "gpt-5-mini", "gpt-5-nano", "gpt-5-1", "gpt-5-2", "gpt-5-4", "gpt-5-4-mini", "gpt-5-4-nano", "gpt-5-5", "gpt-oss-120b", "gpt-oss-20b", "gemini-3-pro", "gemini-3-flash", "gemini-3-1-pro", "gemini-3-1-flash-lite", "gemini-2-5-pro", "gemini-2-5-flash"];
-/**
- * Ids the gateway serves that the models.dev `neon` provider does not list.
- * Verified to resolve against the gateway; retained for autocomplete. If
- * models.dev later lists one of these, the drift check flags it for promotion
- * into `NEON_MODELS_DEV_IDS`.
- */
-declare const NEON_EXTRA_MODEL_IDS: readonly ["claude-opus-4-8", "gpt-5-2-codex", "gpt-5-3-codex", "gpt-5-5-pro", "gemini-3-5-flash", "gemma-3-12b", "llama-4-maverick", "meta-llama-3-3-70b-instruct", "meta-llama-3-1-8b-instruct", "qwen3-next-80b-a3b-instruct", "qwen35-122b-a10b"];
-/** A known Neon AI Gateway model id (models.dev catalog + gateway extras). */
-type NeonKnownModelId = (typeof NEON_MODELS_DEV_IDS)[number] | (typeof NEON_EXTRA_MODEL_IDS)[number];
+declare const NEON_MODELS_DEV_IDS: readonly ["claude-fable-5", "claude-haiku-4-5", "claude-opus-4-1", "claude-opus-4-5", "claude-opus-4-6", "claude-opus-4-7", "claude-opus-4-8", "claude-opus-5", "claude-sonnet-4-5", "claude-sonnet-4-6", "claude-sonnet-5", "gpt-5", "gpt-5-1", "gpt-5-2", "gpt-5-3-codex", "gpt-5-4", "gpt-5-4-mini", "gpt-5-4-nano", "gpt-5-5", "gpt-5-5-pro", "gpt-5-6-luna", "gpt-5-6-sol", "gpt-5-6-terra", "gpt-5-mini", "gpt-5-nano", "gpt-oss-120b", "gpt-oss-20b", "gemini-3-1-flash-lite", "gemini-3-1-pro", "gemini-3-5-flash", "gemini-3-flash", "gemma-3-12b", "llama-4-maverick", "meta-llama-3-1-8b-instruct", "meta-llama-3-3-70b-instruct", "qwen3-next-80b-a3b-instruct", "qwen35-122b-a10b", "glm-5-2", "inkling"];
+/** A known Neon AI Gateway model id. */
+type NeonKnownModelId = (typeof NEON_MODELS_DEV_IDS)[number];
 /**
  * A Neon AI Gateway model id. Known ids are listed for autocomplete; any other
  * id (e.g. a brand-new model, or the legacy `databricks-` prefixed form) is
@@ -17,5 +10,5 @@ type NeonKnownModelId = (typeof NEON_MODELS_DEV_IDS)[number] | (typeof NEON_EXTR
  */
 type NeonChatModelId = NeonKnownModelId | (string & {});
 //#endregion
-export { NEON_EXTRA_MODEL_IDS, NEON_MODELS_DEV_IDS, NeonChatModelId, NeonKnownModelId };
+export { NEON_MODELS_DEV_IDS, NeonChatModelId, NeonKnownModelId };
 //# sourceMappingURL=neon-chat-options.d.ts.map

package/dist/lib/neon-chat-options.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"neon-chat-options.d.ts","names":[],"sources":["../../src/lib/neon-chat-options.ts"],"mappings":";;AAiBa,cAAA,mBA8BH,EAAA,SAAA,CAAA,iBAAA,EAAA,iBAAA,EAAA,iBAAA,EAAA,iBAAA,EAAA,mBAAA,EAAA,mBAAA,EAAA,iBAAA,EAAA,kBAAA,EAAA,OAAA,EAAA,YAAA,EAAA,YAAA,EAAA,SAAA,EAAA,SAAA,EAAA,SAAA,EAAA,cAAA,EAAA,cAAA,EAAA,SAAA,EAAA,cAAA,EAAA,aAAA,EAAA,cAAA,EAAA,gBAAA,EAAA,gBAAA,EAAA,uBAAA,EAAA,gBAAA,EAAA,kBAAA,CAAA;AAQV;AAoBA;AAA4B;AACjB;AACA;AAAoB;AAOnB,cA7BC,oBA6BiB,EAAA,SAAgB,CAAA,iBAAA,EAAA,eAAA,EAAA,eAAA,EAAA,aAAA,EAAA,kBAAA,EAAA,aAAA,EAAA,kBAAA,EAAA,6BAAA,EAAA,4BAAA,EAAA,6BAAA,EAAA,kBAAA,CAAA;;KATlC,gBAAA,WACD,uCACA;;;;;;KAOC,eAAA,GAAkB"}
+{"version":3,"file":"neon-chat-options.d.ts","names":[],"sources":["../../src/lib/neon-chat-options.ts"],"mappings":";;AAsBa,cAAA,mBAgDH,EAAA,SAAA,CAAA,gBAAA,EAAA,kBAAA,EAAA,iBAAA,EAAA,iBAAA,EAAA,iBAAA,EAAA,iBAAA,EAAA,iBAAA,EAAA,eAAA,EAAA,mBAAA,EAAA,mBAAA,EAAA,iBAAA,EAAA,OAAA,EAAA,SAAA,EAAA,SAAA,EAAA,eAAA,EAAA,SAAA,EAAA,cAAA,EAAA,cAAA,EAAA,SAAA,EAAA,aAAA,EAAA,cAAA,EAAA,aAAA,EAAA,eAAA,EAAA,YAAA,EAAA,YAAA,EAAA,cAAA,EAAA,aAAA,EAAA,uBAAA,EAAA,gBAAA,EAAA,kBAAA,EAAA,gBAAA,EAAA,aAAA,EAAA,kBAAA,EAAA,4BAAA,EAAA,6BAAA,EAAA,6BAAA,EAAA,kBAAA,EAAA,SAAA,EAAA,SAAA,CAAA;AAGV;AAOY,KAPA,gBAAA,GAOe,CAAG,OAPS,mBAOO,CAAA,CAAA,MAAA,CAAA;;;;;;KAAlC,eAAA,GAAkB"}

package/dist/lib/neon-chat-options.js

@@ -1,52 +1,47 @@
 //#region src/lib/neon-chat-options.ts
 /** The models.dev `neon` catalog (canonical, unprefixed ids). */
 const NEON_MODELS_DEV_IDS = [
-	"claude-opus-4-7",
-	"claude-opus-4-6",
-	"claude-opus-4-5",
+	"claude-fable-5",
+	"claude-haiku-4-5",
 	"claude-opus-4-1",
-	"claude-sonnet-4-6",
+	"claude-opus-4-5",
+	"claude-opus-4-6",
+	"claude-opus-4-7",
+	"claude-opus-4-8",
+	"claude-opus-5",
 	"claude-sonnet-4-5",
-	"claude-sonnet-4",
-	"claude-haiku-4-5",
+	"claude-sonnet-4-6",
+	"claude-sonnet-5",
 	"gpt-5",
-	"gpt-5-mini",
-	"gpt-5-nano",
 	"gpt-5-1",
 	"gpt-5-2",
+	"gpt-5-3-codex",
 	"gpt-5-4",
 	"gpt-5-4-mini",
 	"gpt-5-4-nano",
 	"gpt-5-5",
+	"gpt-5-5-pro",
+	"gpt-5-6-luna",
+	"gpt-5-6-sol",
+	"gpt-5-6-terra",
+	"gpt-5-mini",
+	"gpt-5-nano",
 	"gpt-oss-120b",
 	"gpt-oss-20b",
-	"gemini-3-pro",
-	"gemini-3-flash",
-	"gemini-3-1-pro",
 	"gemini-3-1-flash-lite",
-	"gemini-2-5-pro",
-	"gemini-2-5-flash"
-];
-/**
-* Ids the gateway serves that the models.dev `neon` provider does not list.
-* Verified to resolve against the gateway; retained for autocomplete. If
-* models.dev later lists one of these, the drift check flags it for promotion
-* into `NEON_MODELS_DEV_IDS`.
-*/
-const NEON_EXTRA_MODEL_IDS = [
-	"claude-opus-4-8",
-	"gpt-5-2-codex",
-	"gpt-5-3-codex",
-	"gpt-5-5-pro",
+	"gemini-3-1-pro",
 	"gemini-3-5-flash",
+	"gemini-3-flash",
 	"gemma-3-12b",
 	"llama-4-maverick",
-	"meta-llama-3-3-70b-instruct",
 	"meta-llama-3-1-8b-instruct",
+	"meta-llama-3-3-70b-instruct",
 	"qwen3-next-80b-a3b-instruct",
-	"qwen35-122b-a10b"
+	"qwen35-122b-a10b",
+	"glm-5-2",
+	"inkling"
 ];
 //#endregion
-export { NEON_EXTRA_MODEL_IDS, NEON_MODELS_DEV_IDS };
+export { NEON_MODELS_DEV_IDS };
 
 //# sourceMappingURL=neon-chat-options.js.map