@ljoukov/llm 4.0.2 → 4.0.4

package/README.md

@@ -333,13 +333,15 @@ Use a `chatgpt-` prefix:
 import { generateText } from "@ljoukov/llm";
 
 const result = await generateText({
-  model: "chatgpt-gpt-5.1-codex-mini",
+  model: "chatgpt-gpt-5.4",
   input: "Return exactly: OK",
 });
 
 console.log(result.text);
 ```
 
+`chatgpt-gpt-5.4-fast` is also supported as a convenience alias for ChatGPT-authenticated `gpt-5.4` with priority processing enabled (`service_tier="priority"`), matching Codex `/fast` semantics.
+
 ## JSON outputs
 
 `generateJson()` validates the output with Zod and returns the parsed value.
@@ -681,7 +683,7 @@ See `docs/agent-telemetry.md` for event schema, design rationale, and backend ad
 
 - console lines,
 - `<workspace>/agent.log`,
-- per-call artifacts under a sibling logs directory: `<workspace-parent>/logs/<timestamp>-<n>/<model-id>/`.
+- per-call artifacts under `<workspace>/llm_calls/<timestamp>-<n>/<model-id>/` by default.
 
 Each LLM call writes:
 
@@ -692,6 +694,7 @@ Each LLM call writes:
 `image_url` data URLs are redacted in text/metadata logs (`data:...,...`) so base64 payloads are not printed inline.
 
 ```ts
+import path from "node:path";
 import { runAgentLoop } from "@ljoukov/llm";
 
 await runAgentLoop({
@@ -699,7 +702,8 @@ await runAgentLoop({
   input: "Do the task",
   filesystemTool: true,
   logging: {
-    workspaceDir: process.cwd(), // optional; defaults to filesystem cwd or process.cwd()
+    workspaceDir: path.join(process.cwd(), "logs", "agent"), // optional; defaults to filesystem cwd or process.cwd()
+    callLogsDir: "llm_calls", // optional; relative paths resolve from workspaceDir
     mirrorToConsole: false, // useful for CLI UIs that already render stream events
     sink: {
       append: (line) => {

package/dist/index.cjs

@@ -317,6 +317,16 @@ var OPENAI_GPT_52_PRICING = {
   cachedRate: 0.175 / 1e6,
   outputRate: 14 / 1e6
 };
+var OPENAI_GPT_54_PRICING = {
+  inputRate: 2.5 / 1e6,
+  cachedRate: 0.25 / 1e6,
+  outputRate: 15 / 1e6
+};
+var OPENAI_GPT_54_PRIORITY_PRICING = {
+  inputRate: 5 / 1e6,
+  cachedRate: 0.5 / 1e6,
+  outputRate: 30 / 1e6
+};
 var OPENAI_GPT_53_CODEX_PRICING = {
   inputRate: 1.25 / 1e6,
   cachedRate: 0.125 / 1e6,
@@ -328,6 +338,12 @@ var OPENAI_GPT_5_MINI_PRICING = {
   outputRate: 2 / 1e6
 };
 function getOpenAiPricing(modelId) {
+  if (modelId.includes("gpt-5.4-fast")) {
+    return OPENAI_GPT_54_PRIORITY_PRICING;
+  }
+  if (modelId.includes("gpt-5.4")) {
+    return OPENAI_GPT_54_PRICING;
+  }
   if (modelId.includes("gpt-5.3-codex-spark")) {
     return OPENAI_GPT_5_MINI_PRICING;
   }
@@ -2738,11 +2754,18 @@ async function runOpenAiCall(fn, modelId, runOptions) {
 }
 
 // src/openai/models.ts
-var OPENAI_MODEL_IDS = ["gpt-5.3-codex", "gpt-5.2", "gpt-5.1-codex-mini"];
+var OPENAI_MODEL_IDS = [
+  "gpt-5.4",
+  "gpt-5.3-codex",
+  "gpt-5.2",
+  "gpt-5.1-codex-mini"
+];
 function isOpenAiModelId(value) {
   return OPENAI_MODEL_IDS.includes(value);
 }
 var CHATGPT_MODEL_IDS = [
+  "chatgpt-gpt-5.4",
+  "chatgpt-gpt-5.4-fast",
   "chatgpt-gpt-5.3-codex",
   "chatgpt-gpt-5.3-codex-spark",
   "chatgpt-gpt-5.2",
@@ -2754,6 +2777,17 @@ function isChatGptModelId(value) {
 function stripChatGptPrefix(model) {
   return model.slice("chatgpt-".length);
 }
+function resolveChatGptProviderModel(model) {
+  switch (model) {
+    case "chatgpt-gpt-5.4-fast":
+      return "gpt-5.4";
+    default:
+      return stripChatGptPrefix(model);
+  }
+}
+function resolveChatGptServiceTier(model) {
+  return model === "chatgpt-gpt-5.4-fast" ? "priority" : void 0;
+}
 
 // src/agentLogging.ts
 var import_node_async_hooks = require("async_hooks");
@@ -2935,7 +2969,8 @@ var AgentLoggingSessionImpl = class {
   callCounter = 0;
   constructor(config) {
     this.workspaceDir = import_node_path3.default.resolve(config.workspaceDir ?? process.cwd());
-    this.logsRootDir = import_node_path3.default.join(import_node_path3.default.dirname(this.workspaceDir), "logs");
+    const configuredCallLogsDir = typeof config.callLogsDir === "string" ? config.callLogsDir.trim() : "";
+    this.logsRootDir = configuredCallLogsDir.length > 0 ? import_node_path3.default.resolve(this.workspaceDir, configuredCallLogsDir) : import_node_path3.default.join(this.workspaceDir, "llm_calls");
     this.mirrorToConsole = config.mirrorToConsole !== false;
     this.sink = config.sink;
     this.agentLogPath = import_node_path3.default.join(this.workspaceDir, "agent.log");
@@ -3502,7 +3537,11 @@ function convertLlmContentToGeminiContent(content) {
 }
 function resolveProvider(model) {
   if (isChatGptModelId(model)) {
-    return { provider: "chatgpt", model: stripChatGptPrefix(model) };
+    return {
+      provider: "chatgpt",
+      model: resolveChatGptProviderModel(model),
+      serviceTier: resolveChatGptServiceTier(model)
+    };
   }
   if (isGeminiTextModelId(model) || isGeminiImageModelId(model)) {
     return { provider: "gemini", model };
@@ -5213,6 +5252,7 @@ async function runTextCall(params) {
         model: modelForProvider,
         store: false,
         stream: true,
+        ...providerInfo.serviceTier ? { service_tier: providerInfo.serviceTier } : {},
         instructions: chatGptInput.instructions ?? "You are a helpful assistant.",
         input: chatGptInput.input,
         include: ["reasoning.encrypted_content"],
@@ -5247,7 +5287,7 @@ async function runTextCall(params) {
         queue.push({ type: "blocked" });
       }
       if (result.model) {
-        modelVersion = `chatgpt-${result.model}`;
+        modelVersion = providerInfo.serviceTier ? request.model : `chatgpt-${result.model}`;
         queue.push({ type: "model", modelVersion });
       }
       latestUsage = extractChatGptUsageTokens(result.usage);
@@ -6209,6 +6249,7 @@ async function runToolLoop(request) {
           model: providerInfo.model,
           store: false,
           stream: true,
+          ...providerInfo.serviceTier ? { service_tier: providerInfo.serviceTier } : {},
           instructions: toolLoopInput.instructions ?? "You are a helpful assistant.",
           input,
           prompt_cache_key: promptCacheKey,
@@ -6249,7 +6290,7 @@ async function runToolLoop(request) {
             }
           });
           const modelCompletedAtMs = Date.now();
-          modelVersion = response.model ? `chatgpt-${response.model}` : request.model;
+          modelVersion = response.model && !providerInfo.serviceTier ? `chatgpt-${response.model}` : request.model;
           usageTokens = extractChatGptUsageTokens(response.usage);
           const stepCostUsd = estimateCallCostUsd({
             modelId: modelVersion,
@@ -10374,9 +10415,10 @@ function createRootAgentLoggingSession(request) {
   if (!selected) {
     return void 0;
   }
+  const workspaceDir = typeof selected.workspaceDir === "string" && selected.workspaceDir.trim().length > 0 ? import_node_path7.default.resolve(selected.workspaceDir) : resolveWorkspaceDirForLogging(request);
   return createAgentLoggingSession({
     ...selected,
-    workspaceDir: typeof selected.workspaceDir === "string" && selected.workspaceDir.trim().length > 0 ? import_node_path7.default.resolve(selected.workspaceDir) : resolveWorkspaceDirForLogging(request),
+    workspaceDir,
     mirrorToConsole: selected.mirrorToConsole !== false
   });
 }