@juspay/neurolink 9.28.0 → 9.29.0

package/CHANGELOG.md

@@ -1,3 +1,15 @@
+## [9.29.0](https://github.com/juspay/neurolink/compare/v9.28.1...v9.29.0) (2026-03-19)
+
+### Features
+
+- **(observability):** add OTEL instrumentation, observability exporters, and comprehensive test suite fixes ([095a774](https://github.com/juspay/neurolink/commit/095a7748b08edac4c34467e2a72caff4426a65bd))
+
+## [9.28.1](https://github.com/juspay/neurolink/compare/v9.28.0...v9.28.1) (2026-03-18)
+
+### Bug Fixes
+
+- **(docs):** comprehensive documentation audit, code example fixes, and model updates ([f294ff2](https://github.com/juspay/neurolink/commit/f294ff2eef71a09e271c5dc536a9d00d97cfd5e5))
+
 ## [9.28.0](https://github.com/juspay/neurolink/compare/v9.27.0...v9.28.0) (2026-03-18)
 
 ### Features

package/README.md

@@ -9,11 +9,14 @@ generated (LLM providers: the neurons) to where they are needed (connectors: the
 ```typescript
 import { NeuroLink } from "@juspay/neurolink";
 
-const pipe = new NeuroLink({ defaultProvider: "anthropic" });
+const pipe = new NeuroLink();
 
 // Everything is a stream
-for await (const token of pipe.stream({ prompt: "Hello" })) {
-  process.stdout.write(token);
+const result = await pipe.stream({ input: { text: "Hello" } });
+for await (const chunk of result.stream) {
+  if ("content" in chunk) {
+    process.stdout.write(chunk.content);
+  }
 }
 ```
 
@@ -79,11 +82,12 @@ Extracted from production systems at Juspay and battle-tested at enterprise scal
 
 ```typescript
 // Image Generation with Gemini (v8.31.0)
-const image = await neurolink.generateImage({
-  prompt: "A futuristic cityscape",
+const image = await neurolink.generate({
+  input: { text: "A futuristic cityscape" },
   provider: "google-ai",
   model: "imagen-3.0-generate-002",
 });
+console.log(image.imageOutput?.base64); // Base64-encoded image
 
 // HTTP Transport for Remote MCP (v8.29.0)
 await neurolink.addExternalMCPServer("remote-tools", {
@@ -398,27 +402,25 @@ import { NeuroLink } from "@juspay/neurolink";
 const neurolink = new NeuroLink({
   conversationMemory: {
     enabled: true,
-    store: "redis", // Automatically uses REDIS_URL
-    ttl: 86400, // 24-hour session expiration
+    enableSummarization: true,
   },
 });
 
-// Or explicit configuration
+// Or explicit Redis configuration
 const neurolinkExplicit = new NeuroLink({
   conversationMemory: {
     enabled: true,
-    store: "redis",
-    redis: {
+    redisConfig: {
       host: "redis.example.com",
       port: 6379,
       password: process.env.REDIS_PASSWORD,
-      tls: true, // Enable for production
+      ttl: 86400, // 24-hour session expiration (seconds)
     },
   },
 });
 
-// Export conversation for analytics
-const history = await neurolink.exportConversation({ format: "json" });
+// Retrieve conversation history for analytics
+const history = await neurolink.getConversationHistory("session-id");
 await saveToDataWarehouse(history);
 ```
 
@@ -605,7 +607,6 @@ import { NeuroLink } from "@juspay/neurolink";
 const neurolink = new NeuroLink({
   conversationMemory: {
     enabled: true,
-    store: "redis",
   },
   enableOrchestration: true,
 });
@@ -633,7 +634,7 @@ console.log(result.evaluation?.overallScore);
 
 // RAG: Ask questions about your documents
 const answer = await neurolink.generate({
-  prompt: "What are the main architectural decisions?",
+  input: { text: "What are the main architectural decisions?" },
   rag: {
     files: ["./docs/architecture.md", "./docs/decisions.md"],
     strategy: "markdown",
@@ -657,7 +658,9 @@ const result = await neurolink.generate({
   },
   provider: "vertex",
   model: "gemini-3-flash-preview",
-  thinkingLevel: "medium", // Options: "minimal", "low", "medium", "high"
+  thinkingConfig: {
+    thinkingLevel: "medium", // Options: "minimal", "low", "medium", "high"
+  },
 });
 
 console.log(result.content);

package/dist/adapters/providerImageAdapter.js

@@ -62,7 +62,12 @@ function normalizeVisionProvider(provider) {
  */
 const VISION_CAPABILITIES = {
     openai: [
-        // GPT-5.2 family (released Dec 11, 2025) - Latest flagship models
+        // GPT-5.4 family (released Mar 2026) - Latest flagship models
+        "gpt-5.4",
+        "gpt-5.4-mini",
+        "gpt-5.4-nano",
+        "gpt-5.4-pro",
+        // GPT-5.2 family (released Dec 11, 2025)
         "gpt-5.2",
         "gpt-5.2-chat-latest",
         "gpt-5.2-pro",
@@ -118,6 +123,9 @@ const VISION_CAPABILITIES = {
         "gemini-pro-vision",
     ],
     anthropic: [
+        // Claude 4.6 Series (February 2026)
+        "claude-opus-4-6",
+        "claude-sonnet-4-6",
         // Claude 4.5 Series (September-November 2025)
         "claude-sonnet-4-5",
         "claude-sonnet-4-5-20250929",

package/dist/constants/contextWindows.js

@@ -25,9 +25,9 @@ export const DEFAULT_OUTPUT_RESERVE_RATIO = 0.35;
 export const MODEL_CONTEXT_WINDOWS = {
     anthropic: {
         _default: 200_000,
-        // Claude 4.6 (Feb 2026) — 200K standard, 1M with beta header
-        "claude-opus-4-6": 200_000,
-        "claude-sonnet-4-6": 200_000,
+        // Claude 4.6 (Feb 2026) — 1M context window
+        "claude-opus-4-6": 1_000_000,
+        "claude-sonnet-4-6": 1_000_000,
         // Claude 4.5
         "claude-opus-4-5-20251101": 200_000,
         "claude-sonnet-4-5-20250929": 200_000,
@@ -46,6 +46,11 @@ export const MODEL_CONTEXT_WINDOWS = {
     },
     openai: {
         _default: 128_000,
+        // GPT-5.4 family — 1.05M context
+        "gpt-5.4": 1_050_000,
+        "gpt-5.4-mini": 400_000,
+        "gpt-5.4-nano": 400_000,
+        "gpt-5.4-pro": 1_050_000,
         // GPT-5.x family — 400K context
         "gpt-5.3-codex": 400_000,
         "gpt-5.2": 400_000,
@@ -108,8 +113,8 @@ export const MODEL_CONTEXT_WINDOWS = {
     vertex: {
         _default: 1_048_576,
         // Claude on Vertex
-        "claude-opus-4-6": 200_000,
-        "claude-sonnet-4-6": 200_000,
+        "claude-opus-4-6": 1_000_000,
+        "claude-sonnet-4-6": 1_000_000,
         "claude-sonnet-4-5": 200_000,
         "claude-opus-4-5": 200_000,
         "claude-haiku-4-5": 200_000,
@@ -136,8 +141,8 @@ export const MODEL_CONTEXT_WINDOWS = {
     bedrock: {
         _default: 200_000,
         // Claude 4.6
-        "anthropic.claude-opus-4-6-v1:0": 200_000,
-        "anthropic.claude-sonnet-4-6": 200_000,
+        "anthropic.claude-opus-4-6-v1:0": 1_000_000,
+        "anthropic.claude-sonnet-4-6": 1_000_000,
         // Claude 4.5
         "anthropic.claude-opus-4-5-20251124-v1:0": 200_000,
         "anthropic.claude-sonnet-4-5-20250929-v1:0": 200_000,
@@ -160,6 +165,11 @@ export const MODEL_CONTEXT_WINDOWS = {
     },
     azure: {
         _default: 128_000,
+        // GPT-5.4
+        "gpt-5.4": 1_050_000,
+        "gpt-5.4-mini": 400_000,
+        "gpt-5.4-nano": 400_000,
+        "gpt-5.4-pro": 1_050_000,
         // GPT-5.x
         "gpt-5.2": 400_000,
         "gpt-5.2-pro": 400_000,
@@ -192,6 +202,7 @@ export const MODEL_CONTEXT_WINDOWS = {
         "devstral-2512": 256_000,
         "devstral-small-2512": 256_000,
         "magistral-medium-latest": 128_000,
+        "mistral-small-2603": 256_000,
     },
     ollama: {
         _default: 128_000,

package/dist/constants/enums.d.ts

@@ -58,10 +58,15 @@ export declare enum BedrockModels {
     CLAUDE_4_5_HAIKU = "anthropic.claude-haiku-4-5-20251001-v1:0",
     CLAUDE_4_1_OPUS = "anthropic.claude-opus-4-1-20250805-v1:0",
     CLAUDE_4_SONNET = "anthropic.claude-sonnet-4-20250514-v1:0",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_7_SONNET = "anthropic.claude-3-7-sonnet-20250219-v1:0",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_5_SONNET = "anthropic.claude-3-5-sonnet-20241022-v1:0",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_5_HAIKU = "anthropic.claude-3-5-haiku-20241022-v1:0",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_SONNET = "anthropic.claude-3-sonnet-20240229-v1:0",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_HAIKU = "anthropic.claude-3-haiku-20240307-v1:0",
     NOVA_PREMIER = "amazon.nova-premier-v1:0",
     NOVA_PRO = "amazon.nova-pro-v1:0",
@@ -144,6 +149,10 @@ export declare enum BedrockModels {
  */
 export declare enum OpenAIModels {
     GPT_5_3_CODEX = "gpt-5.3-codex",
+    GPT_5_4 = "gpt-5.4",
+    GPT_5_4_MINI = "gpt-5.4-mini",
+    GPT_5_4_NANO = "gpt-5.4-nano",
+    GPT_5_4_PRO = "gpt-5.4-pro",
     GPT_5_2 = "gpt-5.2",
     GPT_5_2_CHAT_LATEST = "gpt-5.2-chat-latest",
     GPT_5_2_PRO = "gpt-5.2-pro",
@@ -171,7 +180,9 @@ export declare enum OpenAIModels {
     O3_PRO = "o3-pro",
     O4_MINI = "o4-mini",
     O1 = "o1",
+    /** @deprecated Turned off Jul 14, 2025. Use GPT_4_1 or O3. */
     O1_PREVIEW = "o1-preview",
+    /** @deprecated Replaced by o3-mini. */
     O1_MINI = "o1-mini",
     GPT_4 = "gpt-4",
     GPT_4_TURBO = "gpt-4-turbo",
@@ -186,6 +197,9 @@ export declare enum AzureOpenAIModels {
     GPT_5_2_CHAT = "gpt-5.2-chat",
     GPT_5_2_PRO = "gpt-5.2-pro",
     GPT_5_2_CODEX = "gpt-5.2-codex",
+    GPT_5_4 = "gpt-5.4",
+    GPT_5_4_MINI = "gpt-5.4-mini",
+    GPT_5_4_NANO = "gpt-5.4-nano",
     GPT_5_1 = "gpt-5.1",
     GPT_5_1_CHAT = "gpt-5.1-chat",
     GPT_5_1_CODEX = "gpt-5.1-codex",
@@ -228,11 +242,17 @@ export declare enum VertexModels {
     CLAUDE_4_5_HAIKU = "claude-haiku-4-5@20251001",
     CLAUDE_4_0_SONNET = "claude-sonnet-4@20250514",
     CLAUDE_4_0_OPUS = "claude-opus-4@20250514",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_7_SONNET = "claude-3-7-sonnet@20250219",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_5_SONNET = "claude-3-5-sonnet-20241022",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_5_HAIKU = "claude-3-5-haiku-20241022",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_SONNET = "claude-3-sonnet-20240229",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_OPUS = "claude-3-opus-20240229",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     CLAUDE_3_HAIKU = "claude-3-haiku-20240307",
     GEMINI_3_1_PRO = "gemini-3.1-pro",
     GEMINI_3_1_PRO_PREVIEW = "gemini-3.1-pro-preview",
@@ -253,9 +273,9 @@ export declare enum VertexModels {
     GEMINI_2_0_FLASH = "gemini-2.0-flash",
     GEMINI_2_0_FLASH_001 = "gemini-2.0-flash-001",
     GEMINI_2_0_FLASH_LITE = "gemini-2.0-flash-lite",
-    /** @deprecated Retired Sep 2025. Use Gemini 2.5+ */
+    /** @deprecated SHUT DOWN. Returns 404. Use GEMINI_2_5_FLASH or newer. */
     GEMINI_1_5_PRO = "gemini-1.5-pro-002",
-    /** @deprecated Retired Sep 2025. Use Gemini 2.5+ */
+    /** @deprecated SHUT DOWN. Returns 404. Use GEMINI_2_5_FLASH or newer. */
     GEMINI_1_5_FLASH = "gemini-1.5-flash-002"
 }
 /**
@@ -277,16 +297,18 @@ export declare enum GoogleAIModels {
     GEMINI_2_5_FLASH_IMAGE = "gemini-2.5-flash-image",
     GEMINI_2_5_FLASH_PREVIEW_TTS = "gemini-2.5-flash-preview-tts",
     GEMINI_2_5_PRO_PREVIEW_TTS = "gemini-2.5-pro-preview-tts",
+    /** @deprecated Retiring June 1, 2026. Use GEMINI_2_5_FLASH instead. */
     GEMINI_2_0_FLASH = "gemini-2.0-flash",
     GEMINI_2_0_FLASH_001 = "gemini-2.0-flash-001",
     GEMINI_2_0_FLASH_LITE = "gemini-2.0-flash-lite",
     GEMINI_2_0_FLASH_IMAGE = "gemini-2.0-flash-preview-image-generation",
-    /** @deprecated Retired Sep 2025. Use Gemini 2.5+ */
+    /** @deprecated SHUT DOWN. Returns 404. Use GEMINI_2_5_FLASH or newer. */
     GEMINI_1_5_PRO = "gemini-1.5-pro",
-    /** @deprecated Retired Sep 2025. Use Gemini 2.5+ */
+    /** @deprecated SHUT DOWN. Returns 404. Use GEMINI_2_5_FLASH or newer. */
     GEMINI_1_5_FLASH = "gemini-1.5-flash",
     GEMINI_EMBEDDING = "gemini-embedding-001",
-    /** @deprecated Shutdown Jan 2026. Use gemini-embedding-001 */
+    GEMINI_EMBEDDING_2_PREVIEW = "gemini-embedding-2-preview",
+    /** @deprecated SHUT DOWN Jan 14, 2026. Use GEMINI_EMBEDDING instead. */
     TEXT_EMBEDDING_004 = "text-embedding-004"
 }
 /**
@@ -301,11 +323,17 @@ export declare enum AnthropicModels {
     CLAUDE_OPUS_4_1 = "claude-opus-4-1-20250805",
     CLAUDE_OPUS_4_0 = "claude-opus-4-20250514",
     CLAUDE_SONNET_4_0 = "claude-sonnet-4-20250514",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     CLAUDE_SONNET_3_7 = "claude-3-7-sonnet-20250219",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     CLAUDE_3_5_SONNET = "claude-3-5-sonnet-20241022",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     CLAUDE_3_5_HAIKU = "claude-3-5-haiku-20241022",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     CLAUDE_3_SONNET = "claude-3-sonnet-20240229",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     CLAUDE_3_OPUS = "claude-3-opus-20240229",
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     CLAUDE_3_HAIKU = "claude-3-haiku-20240307"
 }
 /**
@@ -341,7 +369,9 @@ export declare enum MistralModels {
     MISTRAL_OCR_LATEST = "mistral-ocr-latest",
     MISTRAL_NEMO = "mistral-nemo",
     MISTRAL_EMBED = "mistral-embed",
-    MISTRAL_MODERATION_LATEST = "mistral-moderation-latest"
+    MISTRAL_MODERATION_LATEST = "mistral-moderation-latest",
+    MISTRAL_SMALL_4 = "mistral-small-2603",
+    MISTRAL_SMALL_CREATIVE = "mistral-small-creative"
 }
 /**
  * Supported Models for Ollama (Local)

package/dist/constants/enums.js

@@ -78,12 +78,17 @@ export var BedrockModels;
     BedrockModels["CLAUDE_4_1_OPUS"] = "anthropic.claude-opus-4-1-20250805-v1:0";
     BedrockModels["CLAUDE_4_SONNET"] = "anthropic.claude-sonnet-4-20250514-v1:0";
     // Claude 3.7 Series
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     BedrockModels["CLAUDE_3_7_SONNET"] = "anthropic.claude-3-7-sonnet-20250219-v1:0";
     // Claude 3.5 Series
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     BedrockModels["CLAUDE_3_5_SONNET"] = "anthropic.claude-3-5-sonnet-20241022-v1:0";
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     BedrockModels["CLAUDE_3_5_HAIKU"] = "anthropic.claude-3-5-haiku-20241022-v1:0";
     // Claude 3 Series (Legacy support)
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     BedrockModels["CLAUDE_3_SONNET"] = "anthropic.claude-3-sonnet-20240229-v1:0";
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     BedrockModels["CLAUDE_3_HAIKU"] = "anthropic.claude-3-haiku-20240307-v1:0";
     // ============================================================================
     // AMAZON NOVA MODELS
@@ -216,6 +221,11 @@ export var OpenAIModels;
 (function (OpenAIModels) {
     // GPT-5.3 Series (Released February 2026) - Latest coding models
     OpenAIModels["GPT_5_3_CODEX"] = "gpt-5.3-codex";
+    // GPT-5.4 Series (Released March 2026) - Latest flagship models
+    OpenAIModels["GPT_5_4"] = "gpt-5.4";
+    OpenAIModels["GPT_5_4_MINI"] = "gpt-5.4-mini";
+    OpenAIModels["GPT_5_4_NANO"] = "gpt-5.4-nano";
+    OpenAIModels["GPT_5_4_PRO"] = "gpt-5.4-pro";
     // GPT-5.2 Series (Released December 11, 2025) - Flagship models
     OpenAIModels["GPT_5_2"] = "gpt-5.2";
     OpenAIModels["GPT_5_2_CHAT_LATEST"] = "gpt-5.2-chat-latest";
@@ -250,7 +260,9 @@ export var OpenAIModels;
     OpenAIModels["O3_PRO"] = "o3-pro";
     OpenAIModels["O4_MINI"] = "o4-mini";
     OpenAIModels["O1"] = "o1";
+    /** @deprecated Turned off Jul 14, 2025. Use GPT_4_1 or O3. */
     OpenAIModels["O1_PREVIEW"] = "o1-preview";
+    /** @deprecated Replaced by o3-mini. */
     OpenAIModels["O1_MINI"] = "o1-mini";
     // GPT-4 Series (Legacy)
     OpenAIModels["GPT_4"] = "gpt-4";
@@ -269,6 +281,10 @@ export var AzureOpenAIModels;
     AzureOpenAIModels["GPT_5_2_CHAT"] = "gpt-5.2-chat";
     AzureOpenAIModels["GPT_5_2_PRO"] = "gpt-5.2-pro";
     AzureOpenAIModels["GPT_5_2_CODEX"] = "gpt-5.2-codex";
+    // GPT-5.4 Series (March 2026)
+    AzureOpenAIModels["GPT_5_4"] = "gpt-5.4";
+    AzureOpenAIModels["GPT_5_4_MINI"] = "gpt-5.4-mini";
+    AzureOpenAIModels["GPT_5_4_NANO"] = "gpt-5.4-nano";
     // GPT-5.1 Series (October 2025)
     AzureOpenAIModels["GPT_5_1"] = "gpt-5.1";
     AzureOpenAIModels["GPT_5_1_CHAT"] = "gpt-5.1-chat";
@@ -323,13 +339,19 @@ export var VertexModels;
     VertexModels["CLAUDE_4_0_SONNET"] = "claude-sonnet-4@20250514";
     VertexModels["CLAUDE_4_0_OPUS"] = "claude-opus-4@20250514";
     // Claude 3.7 Series (February 2025)
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     VertexModels["CLAUDE_3_7_SONNET"] = "claude-3-7-sonnet@20250219";
     // Claude 3.5 Series (Still supported)
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     VertexModels["CLAUDE_3_5_SONNET"] = "claude-3-5-sonnet-20241022";
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     VertexModels["CLAUDE_3_5_HAIKU"] = "claude-3-5-haiku-20241022";
     // Claude 3 Series (Legacy support)
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     VertexModels["CLAUDE_3_SONNET"] = "claude-3-sonnet-20240229";
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     VertexModels["CLAUDE_3_OPUS"] = "claude-3-opus-20240229";
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_4_6_SONNET instead. */
     VertexModels["CLAUDE_3_HAIKU"] = "claude-3-haiku-20240307";
     // Gemini 3.1 Series (Released March 2026)
     VertexModels["GEMINI_3_1_PRO"] = "gemini-3.1-pro";
@@ -355,9 +377,9 @@ export var VertexModels;
     VertexModels["GEMINI_2_0_FLASH_001"] = "gemini-2.0-flash-001";
     VertexModels["GEMINI_2_0_FLASH_LITE"] = "gemini-2.0-flash-lite";
     // Gemini 1.5 Series (Retired - returns 404)
-    /** @deprecated Retired Sep 2025. Use Gemini 2.5+ */
+    /** @deprecated SHUT DOWN. Returns 404. Use GEMINI_2_5_FLASH or newer. */
     VertexModels["GEMINI_1_5_PRO"] = "gemini-1.5-pro-002";
-    /** @deprecated Retired Sep 2025. Use Gemini 2.5+ */
+    /** @deprecated SHUT DOWN. Returns 404. Use GEMINI_2_5_FLASH or newer. */
     VertexModels["GEMINI_1_5_FLASH"] = "gemini-1.5-flash-002";
 })(VertexModels || (VertexModels = {}));
 /**
@@ -384,18 +406,20 @@ export var GoogleAIModels;
     GoogleAIModels["GEMINI_2_5_FLASH_PREVIEW_TTS"] = "gemini-2.5-flash-preview-tts";
     GoogleAIModels["GEMINI_2_5_PRO_PREVIEW_TTS"] = "gemini-2.5-pro-preview-tts";
     // Gemini 2.0 Series (Deprecated - retiring Jun 2026)
+    /** @deprecated Retiring June 1, 2026. Use GEMINI_2_5_FLASH instead. */
     GoogleAIModels["GEMINI_2_0_FLASH"] = "gemini-2.0-flash";
     GoogleAIModels["GEMINI_2_0_FLASH_001"] = "gemini-2.0-flash-001";
     GoogleAIModels["GEMINI_2_0_FLASH_LITE"] = "gemini-2.0-flash-lite";
     GoogleAIModels["GEMINI_2_0_FLASH_IMAGE"] = "gemini-2.0-flash-preview-image-generation";
     // Gemini 1.5 Series (Retired - returns 404)
-    /** @deprecated Retired Sep 2025. Use Gemini 2.5+ */
+    /** @deprecated SHUT DOWN. Returns 404. Use GEMINI_2_5_FLASH or newer. */
     GoogleAIModels["GEMINI_1_5_PRO"] = "gemini-1.5-pro";
-    /** @deprecated Retired Sep 2025. Use Gemini 2.5+ */
+    /** @deprecated SHUT DOWN. Returns 404. Use GEMINI_2_5_FLASH or newer. */
     GoogleAIModels["GEMINI_1_5_FLASH"] = "gemini-1.5-flash";
     // Embedding Models
     GoogleAIModels["GEMINI_EMBEDDING"] = "gemini-embedding-001";
-    /** @deprecated Shutdown Jan 2026. Use gemini-embedding-001 */
+    GoogleAIModels["GEMINI_EMBEDDING_2_PREVIEW"] = "gemini-embedding-2-preview";
+    /** @deprecated SHUT DOWN Jan 14, 2026. Use GEMINI_EMBEDDING instead. */
     GoogleAIModels["TEXT_EMBEDDING_004"] = "text-embedding-004";
 })(GoogleAIModels || (GoogleAIModels = {}));
 /**
@@ -416,13 +440,19 @@ export var AnthropicModels;
     AnthropicModels["CLAUDE_OPUS_4_0"] = "claude-opus-4-20250514";
     AnthropicModels["CLAUDE_SONNET_4_0"] = "claude-sonnet-4-20250514";
     // Claude 3.7 Series (Legacy)
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     AnthropicModels["CLAUDE_SONNET_3_7"] = "claude-3-7-sonnet-20250219";
     // Claude 3.5 Series (Legacy)
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     AnthropicModels["CLAUDE_3_5_SONNET"] = "claude-3-5-sonnet-20241022";
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     AnthropicModels["CLAUDE_3_5_HAIKU"] = "claude-3-5-haiku-20241022";
     // Claude 3 Series (Legacy - Deprecated)
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     AnthropicModels["CLAUDE_3_SONNET"] = "claude-3-sonnet-20240229";
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     AnthropicModels["CLAUDE_3_OPUS"] = "claude-3-opus-20240229";
+    /** @deprecated Retired from Anthropic API. Use CLAUDE_SONNET_4_6 instead. */
     AnthropicModels["CLAUDE_3_HAIKU"] = "claude-3-haiku-20240307";
 })(AnthropicModels || (AnthropicModels = {}));
 /**
@@ -474,6 +504,9 @@ export var MistralModels;
     MistralModels["MISTRAL_NEMO"] = "mistral-nemo";
     MistralModels["MISTRAL_EMBED"] = "mistral-embed";
     MistralModels["MISTRAL_MODERATION_LATEST"] = "mistral-moderation-latest";
+    // Mistral Small 4 Series (June 2026)
+    MistralModels["MISTRAL_SMALL_4"] = "mistral-small-2603";
+    MistralModels["MISTRAL_SMALL_CREATIVE"] = "mistral-small-creative";
 })(MistralModels || (MistralModels = {}));
 /**
  * Supported Models for Ollama (Local)

package/dist/constants/tokens.d.ts

@@ -71,6 +71,10 @@ export declare const PROVIDER_TOKEN_LIMITS: {
     };
     /** OpenAI model limits */
     readonly OPENAI: {
+        readonly "gpt-5.4": 128000;
+        readonly "gpt-5.4-mini": 128000;
+        readonly "gpt-5.4-nano": 128000;
+        readonly "gpt-5.4-pro": 128000;
         readonly "gpt-4o": 16384;
         readonly "gpt-4o-mini": 16384;
         readonly "gpt-3.5-turbo": 4096;

package/dist/constants/tokens.js

@@ -74,6 +74,10 @@ export const PROVIDER_TOKEN_LIMITS = {
     },
     /** OpenAI model limits */
     OPENAI: {
+        "gpt-5.4": 128_000,
+        "gpt-5.4-mini": 128_000,
+        "gpt-5.4-nano": 128_000,
+        "gpt-5.4-pro": 128_000,
         "gpt-4o": 16384,
         "gpt-4o-mini": 16384,
         "gpt-3.5-turbo": 4096,

package/dist/context/contextCompactor.js

@@ -40,7 +40,7 @@ export class ContextCompactor {
      * Run the multi-stage compaction pipeline until messages fit within budget.
      */
     async compact(messages, targetTokens, memoryConfig, requestId) {
-        const span = SpanSerializer.createSpan(SpanType.CONTEXT_COMPACTION, "context.compact", {
+        let span = SpanSerializer.createSpan(SpanType.CONTEXT_COMPACTION, "context.compact", {
             "context.operation": "compact",
             "context.targetTokens": targetTokens,
         });
@@ -132,7 +132,7 @@ export class ContextCompactor {
                         saved: 0,
                     });
                     // Record failure on the compaction span for trace visibility
-                    SpanSerializer.updateAttributes(span, {
+                    span = SpanSerializer.updateAttributes(span, {
                         "compaction.stage3.error": err.message,
                         "compaction.stage3.errorName": err.name,
                         "compaction.stage3.tokensBefore": stageTokensBefore,

package/dist/core/baseProvider.d.ts

@@ -24,11 +24,15 @@ export declare abstract class BaseProvider implements AIProvider {
     protected sessionId?: string;
     protected userId?: string;
     protected neurolink?: NeuroLink;
-    /** Trace context propagated from NeuroLink SDK for span hierarchy */
-    _traceContext: {
+    /** @internal Trace context propagated from NeuroLink SDK for span hierarchy */
+    protected _traceContext: {
         traceId: string;
         parentSpanId: string;
     } | null;
+    setTraceContext(ctx: {
+        traceId: string;
+        parentSpanId: string;
+    } | null): void;
     private readonly messageBuilder;
     private readonly streamHandler;
     private readonly generationHandler;

package/dist/core/baseProvider.js

@@ -3,7 +3,10 @@ import { generateText } from "ai";
 import { directAgentTools } from "../agent/directTools.js";
 import { IMAGE_GENERATION_MODELS } from "../core/constants.js";
 import { MiddlewareFactory } from "../middleware/factory.js";
+import { SpanStatus, SpanType } from "../observability/types/spanTypes.js";
+import { SpanSerializer } from "../observability/utils/spanSerializer.js";
 import { ATTR, tracers } from "../telemetry/index.js";
+import { calculateCost } from "../utils/pricing.js";
 import { isAbortError } from "../utils/errorHandling.js";
 import { logger } from "../utils/logger.js";
 import { composeAbortSignals, createTimeoutController, TimeoutError, } from "../utils/timeout.js";
@@ -37,8 +40,11 @@ export class BaseProvider {
     sessionId;
     userId;
     neurolink; // Reference to actual NeuroLink instance for MCP tools
-    /** Trace context propagated from NeuroLink SDK for span hierarchy */
+    /** @internal Trace context propagated from NeuroLink SDK for span hierarchy */
     _traceContext = null;
+    setTraceContext(ctx) {
+        this._traceContext = ctx;
+    }
     // Composition modules - Single Responsibility Principle
     messageBuilder;
     streamHandler;
@@ -81,6 +87,14 @@ export class BaseProvider {
      */
     async stream(optionsOrPrompt, analysisSchema) {
         let options = this.normalizeStreamOptions(optionsOrPrompt);
+        // Observability: create metrics span for provider.stream
+        const metricsSpan = SpanSerializer.createSpan(SpanType.MODEL_GENERATION, "provider.stream", {
+            "ai.provider": this.providerName || "unknown",
+            "ai.model": this.modelName || options.model || "unknown",
+            "ai.temperature": options.temperature,
+            "ai.max_tokens": options.maxTokens,
+        }, this._traceContext?.parentSpanId, this._traceContext?.traceId);
+        let metricsSpanRecorded = false;
         // OTEL span for provider-level stream tracing
         const otelStreamSpan = tracers.provider.startSpan("neurolink.provider.stream", {
             kind: SpanKind.CLIENT,
@@ -173,6 +187,11 @@ export class BaseProvider {
             }
         }
         catch (error) {
+            // Observability: record failed stream span
+            metricsSpanRecorded = true;
+            const _endedStreamSpan = SpanSerializer.endSpan(metricsSpan, SpanStatus.ERROR, error instanceof Error ? error.message : String(error));
+            // Note: Do NOT record to getMetricsAggregator() here — neurolink.ts
+            // stream:complete listener handles authoritative metrics to avoid double-counting.
             otelStreamSpan.setStatus({
                 code: SpanStatusCode.ERROR,
                 message: error instanceof Error ? error.message : String(error),
@@ -181,6 +200,12 @@ export class BaseProvider {
             throw error;
         }
         finally {
+            // Observability: record successful stream span (only if not already ended via error path)
+            if (!metricsSpanRecorded) {
+                const _endedStreamSpan = SpanSerializer.endSpan(metricsSpan, SpanStatus.OK);
+                // Note: Do NOT record to getMetricsAggregator() here — neurolink.ts
+                // stream:complete listener handles authoritative metrics to avoid double-counting.
+            }
             // End OTEL span on success (only if not already ended via error path)
             if (otelStreamSpan.isRecording()) {
                 otelStreamSpan.setStatus({ code: SpanStatusCode.OK });
@@ -477,6 +502,13 @@ export class BaseProvider {
         const options = this.normalizeTextOptions(optionsOrPrompt);
         this.validateOptions(options);
         const startTime = Date.now();
+        // Observability: create metrics span for provider.generate
+        const metricsSpan = SpanSerializer.createSpan(SpanType.MODEL_GENERATION, "provider.generate", {
+            "ai.provider": this.providerName || "unknown",
+            "ai.model": this.modelName || options.model || "unknown",
+            "ai.temperature": options.temperature,
+            "ai.max_tokens": options.maxTokens,
+        }, this._traceContext?.parentSpanId, this._traceContext?.traceId);
         // OTEL span for provider-level generate tracing
         // Use startActiveSpan pattern via context.with() so child spans become descendants
         const otelSpan = tracers.provider.startSpan("neurolink.provider.generate", {
@@ -665,9 +697,37 @@ export class BaseProvider {
                         });
                     }
                 }
+                // Observability: record successful generate span with token/cost data
+                let enrichedGenerateSpan = { ...metricsSpan };
+                if (enhancedResult?.usage) {
+                    enrichedGenerateSpan = SpanSerializer.enrichWithTokenUsage(enrichedGenerateSpan, {
+                        promptTokens: enhancedResult.usage.input || 0,
+                        completionTokens: enhancedResult.usage.output || 0,
+                        totalTokens: enhancedResult.usage.total || 0,
+                    });
+                    const cost = calculateCost(this.providerName, this.modelName, {
+                        input: enhancedResult.usage.input || 0,
+                        output: enhancedResult.usage.output || 0,
+                        total: enhancedResult.usage.total || 0,
+                    });
+                    if (cost && cost > 0) {
+                        enrichedGenerateSpan = SpanSerializer.enrichWithCost(enrichedGenerateSpan, {
+                            totalCost: cost,
+                        });
+                    }
+                }
+                const _endedGenerateSpan = SpanSerializer.endSpan(enrichedGenerateSpan, SpanStatus.OK);
+                // Note: Do NOT record to getMetricsAggregator() here — the neurolink.ts
+                // generation:end listener creates an authoritative span with richer context
+                // (provider name, model, input/output) and records to both aggregators.
+                // Recording here would double-count cost and token metrics.
                 return await this.enhanceResult(enhancedResult, options, startTime);
             }
             catch (error) {
+                // Observability: record failed generate span
+                const _endedGenerateSpan = SpanSerializer.endSpan(metricsSpan, SpanStatus.ERROR, error instanceof Error ? error.message : String(error));
+                // Note: Do NOT record to getMetricsAggregator() here — neurolink.ts
+                // handles authoritative metrics recording to avoid double-counting.
                 otelSpan.setStatus({
                     code: SpanStatusCode.ERROR,
                     message: error instanceof Error ? error.message : String(error),