repospend 0.1.5 → 0.1.7

package/CHANGELOG.md

@@ -2,6 +2,25 @@
 
 All notable changes to RepoSpend will be documented in this file.
 
+## Unreleased
+
+## 0.1.7
+
+- Correct Claude Sonnet 5.5 cache reads to $0.10 per million tokens following the October 7 price reduction, without changing Sonnet 5 or intentional custom overrides.
+- Add Claude Haiku 5.5 pricing for prompts up to and over 100,000 tokens, including both cache-write TTLs. Price deduplicated Claude and Copilot inference requests before summing session costs; leave aggregate-only Haiku usage unpriced when request lengths are unknown.
+- Add Gemini 4 Argon and Copilot aliases at introductory $2 input / $0.10 cached input / $10 output rates, with the November 30 promotion expiry recorded in pricing documentation.
+- Preserve exact custom snapshot cards and migrate unchanged Sonnet 5.5 rows in legacy full-table pricing settings.
+- Sum unrounded request costs before rounding the session estimate to six decimal places.
+- Keep inherited Haiku snapshot edits on the shared base card, expose the higher-tier card alongside used models, and explain missing request context and auxiliary model rates in pricing triage.
+- Preserve Haiku rate-card identity while typing or clearing numeric fields, keep pricing columns compact, and show readable request-pricing diagnostics across repo views.
+- Cache model resolution within each session's cost calculation and preserve tolerated numeric-string Claude input counts during deduplication.
+
+## 0.1.6
+
+- Add explicit GPT-6.1 Sol Standard API-equivalent pricing, including its lower $0.10 cached-input rate, without changing GPT-6 Sol estimates.
+- Resolve date-suffixed GPT model IDs to the correct generation while preserving custom snapshot overrides, including case-insensitive and provider-prefixed aliases, and unknown fast-mode cards.
+- Add model pricing and Codex usage guide links to the README and AI-readable documentation, with current pricing sources and estimate limits.
+
 ## 0.1.5
 
 - Correct GPT-5.6 Sol and its alias to the current promotional rates, preserving custom pricing overrides.

package/README.md

@@ -1,4 +1,4 @@
-# RepoSpend
+# RepoSpend — local AI coding cost tracker
 
 RepoSpend is a local-first dashboard for tracking AI coding token usage and API-equivalent spend by repository, session, model, and tool. It supports local Codex, Claude Code, and GitHub Copilot usage data, runs with `npx repospend`, and does not upload prompts, code, transcripts, or usage data.
 
@@ -19,6 +19,26 @@ Project links:
 - npm package: [`repospend`](https://www.npmjs.com/package/repospend)
 - AI-readable summary: [docs/llms.txt](docs/llms.txt)
 
+## Model pricing and usage guides
+
+The 0.1.7 pricing update (October 17) adds Claude Haiku 5.5's two prompt-length tiers,
+reduces Sonnet 5.5 cache reads to $0.10 per million tokens, and adds Gemini 4
+Argon's introductory rates through November 30. Claude and Copilot inference
+records are priced per request; aggregate-only Haiku usage stays unpriced when
+request lengths are unavailable. These are current API-equivalent estimates,
+not historical invoices. See [pricing methodology](docs/pricing.md) for rates,
+promotion dates, and custom overrides.
+
+RepoSpend includes explicit **GPT-6.1 Sol** API-equivalent pricing: $2 input,
+$0.10 cached input, $2.50 cache writes, and $10 output per million tokens at
+Standard short-context rates. GPT-6 Sol keeps its separate $0.20 cache-read rate.
+Estimates describe local token usage, not your ChatGPT or Codex subscription bill.
+
+- [GPT-6.1 Sol pricing and Codex cost tracking](https://repospend.com/guides/gpt-6-1-sol-pricing) — caching, a worked estimate, and local repository usage.
+- [The AI coding price war: Claude 5.5 vs GPT-6.1 Sol](https://repospend.com/guides/ai-coding-price-war) — provider rates and cost per successful task.
+- [Pricing methodology and local overrides](docs/pricing.md)
+- [Token accounting: input, caching, output, and reasoning](docs/token-accounting.md)
+
 ```bash
 npx repospend
 ```

package/dist/cli.js

@@ -253,9 +253,11 @@ function claudePricingFamilyModel(model, isUsable = () => true) {
     "claude-sonnet-4-5",
     "claude-sonnet-4",
     "claude-haiku-4-5",
+    "claude-haiku-5-5",
     "claude-3-5-haiku"
   ];
   const normalized = normalizePricingModelId(model);
+  if (normalized.includes("-long-context")) return void 0;
   const matchesFamily = (candidate) => normalized === candidate || normalized.startsWith(`${candidate}-`) || normalized.startsWith(`${candidate}.`);
   if (normalized.includes("fast-mode")) {
     return families.find((candidate) => candidate.includes("fast-mode") && matchesFamily(candidate) && isUsable(candidate));
@@ -271,7 +273,7 @@ function copilotPricingFamilyModel(model, isUsable = () => true) {
   if ((normalized === "gemini-3-flash" || normalized.startsWith("gemini-3-flash-")) && isUsable("gemini-3-flash")) return "gemini-3-flash";
   if ((normalized === "gemini-3.1-pro" || normalized.startsWith("gemini-3.1-pro-")) && isUsable("gemini-3.1-pro")) return "gemini-3.1-pro";
   if ((normalized === "gemini-3.5-flash" || normalized.startsWith("gemini-3.5-flash-")) && isUsable("gemini-3.5-flash")) return "gemini-3.5-flash";
-  for (const candidate of ["gemini-3.6-flash", "gemini-3.7-flash", "gemini-3.8-flash", "grok-4.5", "grok-4.6", "grok-4.7", "mai-code-1.1-flash", "kimi-k3"]) {
+  for (const candidate of ["gemini-3.6-flash", "gemini-3.7-flash", "gemini-3.8-flash", "gemini-4-argon", "grok-4.5", "grok-4.6", "grok-4.7", "mai-code-1.1-flash", "kimi-k3"]) {
     if ((normalized === candidate || normalized.startsWith(`${candidate}-`)) && isUsable(candidate)) return candidate;
   }
   if ((normalized === "raptor-mini" || normalized.startsWith("raptor-mini-") || normalized.startsWith("oswe-vscode")) && isUsable("raptor-mini")) return "raptor-mini";
@@ -282,13 +284,14 @@ function copilotPricingFamilyModel(model, isUsable = () => true) {
   return void 0;
 }
 function parseVersionedModel(normalized) {
-  if (normalized.includes("fast-mode")) return void 0;
+  if (normalized.includes("fast-mode") || normalized.includes("-long-context")) return void 0;
   const claude = normalized.match(/^(claude-(?:fable|mythos|opus|sonnet|haiku))-(\d+)(?:-(\d{1,2}))?(?!\d)/);
   if (claude) {
     const minor = claude[3] ? Number(claude[3]) : 0;
     return { tier: claude[1], version: Number(claude[2]) + minor / 100 };
   }
-  const gpt = normalized.match(/^(gpt)-(\d)(?:\.(\d+))?(-[a-z][a-z0-9-]*)?$/);
+  const gptModel = normalized.replace(/-(?:\d{4}-\d{2}-\d{2}|\d{8})$/, "");
+  const gpt = gptModel.match(/^(gpt)-(\d)(?:\.(\d+))?(-[a-z][a-z0-9-]*)?$/);
   if (gpt) {
     const minor = gpt[3] ? Number(gpt[3]) : 0;
     return { tier: `gpt${gpt[4] ?? ""}`, version: Number(gpt[2]) + minor / 100 };
@@ -317,17 +320,36 @@ function positiveRate(value) {
 function isUsableModelPricing(pricing) {
   return positiveRate(pricing?.inputPerMillion) && positiveRate(pricing?.outputPerMillion);
 }
-function findModelPricing(model, pricing) {
+function resolvePricingForModel(model, table) {
   const lower = normalizePricingModelId(model);
-  const exact = pricing[model] ?? pricing[lower];
-  if (isUsableModelPricing(exact)) return exact;
-  const versioned = versionedFallbackModel(lower, Object.keys(pricing), (candidate) => isUsableModelPricing(pricing[candidate]));
-  if (versioned) return pricing[versioned];
-  const claudeFamily = claudePricingFamilyModel(lower, (candidate) => isUsableModelPricing(pricing[candidate]));
-  if (claudeFamily) return pricing[claudeFamily];
-  const copilotFamily = copilotPricingFamilyModel(lower, (candidate) => isUsableModelPricing(pricing[candidate]));
-  if (copilotFamily) return pricing[copilotFamily];
-  return void 0;
+  const exactModel = table[model] ? model : table[lower] ? lower : void 0;
+  const exactPricing = exactModel ? table[exactModel] : void 0;
+  if (isUsableModelPricing(exactPricing)) {
+    return { pricing: exactPricing, sourceModel: exactModel, inherited: false };
+  }
+  const versioned = versionedFallbackModel(lower, Object.keys(table), (candidate) => isUsableModelPricing(table[candidate]));
+  if (versioned) {
+    return { pricing: table[versioned], sourceModel: versioned, inherited: versioned !== model && versioned !== lower };
+  }
+  const claudeFamily = claudePricingFamilyModel(lower, (candidate) => isUsableModelPricing(table[candidate]));
+  if (claudeFamily) {
+    return { pricing: table[claudeFamily], sourceModel: claudeFamily, inherited: claudeFamily !== model && claudeFamily !== lower };
+  }
+  const copilotFamily = copilotPricingFamilyModel(lower, (candidate) => isUsableModelPricing(table[candidate]));
+  if (copilotFamily) {
+    return { pricing: table[copilotFamily], sourceModel: copilotFamily, inherited: copilotFamily !== model && copilotFamily !== lower };
+  }
+  return {
+    pricing: exactPricing ?? {
+      inputPerMillion: 0,
+      cacheCreationInputPerMillion: 0,
+      cachedInputPerMillion: 0,
+      outputPerMillion: 0,
+      reasoningOutputPerMillion: 0
+    },
+    sourceModel: exactModel,
+    inherited: false
+  };
 }
 
 // packages/core/src/pricing.ts
@@ -337,18 +359,22 @@ var pricingInfo = {
   sourceUrls: [
     { label: "OpenAI pricing reference", url: "https://developers.openai.com/api/docs/pricing" },
     { label: "OpenAI GPT-6 model reference", url: "https://developers.openai.com/api/docs/models" },
+    { label: "OpenAI GPT-6.1 Sol reference", url: "https://developers.openai.com/api/docs/models/gpt-6.1-sol" },
     { label: "OpenAI GPT-5.6 preview pricing", url: "https://openai.com/index/previewing-gpt-5-6-sol/" },
     { label: "OpenAI GPT-5.6 price update", url: "https://openai.com/index/advancing-the-price-performance-frontier-with-gpt-5-6/" },
     { label: "Claude pricing reference", url: "https://platform.claude.com/docs/en/about-claude/pricing" },
+    { label: "Claude Haiku 5.5 and Sonnet cache update", url: "https://www.anthropic.com/claude-haiku-5-5" },
+    { label: "Gemini 4 Argon pricing", url: "https://ai.google.dev/gemini-api/docs/latest-model?hl=en" },
     { label: "GitHub Copilot model pricing reference", url: "https://docs.github.com/en/copilot/reference/copilot-billing/models-and-pricing" }
   ],
   unit: "USD per 1M tokens",
-  updatedAt: "2026-09-28",
+  updatedAt: "2026-10-17",
   note: "RepoSpend estimates API-equivalent cost from local token counts and public Standard pricing. GPT-6 defaults use short-context rates. Claude transcript fast-mode speeds use explicit published cards. Long-context, other processing modes, regional, and account-specific pricing can differ. This is not your actual bill."
 };
 var defaultPricing = {
   "gpt-6-astra": { inputPerMillion: 10, cachedInputPerMillion: 1, cacheCreationInputPerMillion: 12.5, outputPerMillion: 50, reasoningOutputPerMillion: 50 },
   "gpt-6-sol": { inputPerMillion: 2, cachedInputPerMillion: 0.2, cacheCreationInputPerMillion: 2.5, outputPerMillion: 10, reasoningOutputPerMillion: 10 },
+  "gpt-6.1-sol": { inputPerMillion: 2, cachedInputPerMillion: 0.1, cacheCreationInputPerMillion: 2.5, outputPerMillion: 10, reasoningOutputPerMillion: 10 },
   "gpt-6-luna": { inputPerMillion: 0.1, cachedInputPerMillion: 0.01, cacheCreationInputPerMillion: 0.125, outputPerMillion: 0.5, reasoningOutputPerMillion: 0.5 },
   "gpt-5.6": { inputPerMillion: 4, cachedInputPerMillion: 0.4, cacheCreationInputPerMillion: 5, outputPerMillion: 20, reasoningOutputPerMillion: 20, note: "GPT-5.6 aliases Sol. Promotional Standard short-context rates are available at least through November 21, 2026: $4 input / $20 output, $0.40 cache reads, and $5 cache writes per 1M tokens." },
   "gpt-5.6-sol": { inputPerMillion: 4, cachedInputPerMillion: 0.4, cacheCreationInputPerMillion: 5, outputPerMillion: 20, reasoningOutputPerMillion: 20, note: "Promotional Standard short-context rates available at least through November 21, 2026." },
@@ -384,6 +410,7 @@ var defaultPricing = {
   "gemini-3.6-flash": { inputPerMillion: 0.75, cachedInputPerMillion: 0.075, outputPerMillion: 3.75, note: "GitHub Copilot promotional rates through December 31, 2026; API-equivalent estimate." },
   "gemini-3.7-flash": { inputPerMillion: 0.75, cachedInputPerMillion: 0.075, outputPerMillion: 3.75, note: "GitHub Copilot promotional rates through December 31, 2026; API-equivalent estimate." },
   "gemini-3.8-flash": { inputPerMillion: 0.75, cachedInputPerMillion: 0.075, outputPerMillion: 3.75, note: "GitHub Copilot promotional rates through December 31, 2026; API-equivalent estimate." },
+  "gemini-4-argon": { inputPerMillion: 2, cachedInputPerMillion: 0.1, outputPerMillion: 10, reasoningOutputPerMillion: 10, note: "Introductory rates through November 30, 2026. From December 1: $4 input / $0.20 cached input / $20 output (including thinking) per 1M tokens. No explicit cache-write charge. Bundled prices are a dated snapshot, not an automatic future rate switch." },
   "grok-4.5": { inputPerMillion: 2, cachedInputPerMillion: 0.5, outputPerMillion: 6, note: "GitHub Copilot Standard rates up to 200K input tokens; longer requests use different rates." },
   "grok-4.6": { inputPerMillion: 2, cachedInputPerMillion: 0.5, outputPerMillion: 6, note: "GitHub Copilot Standard rates up to 200K input tokens; longer requests use different rates." },
   "grok-4.7": { inputPerMillion: 2, cachedInputPerMillion: 0.5, outputPerMillion: 6, note: "GitHub Copilot Standard rates up to 200K input tokens; longer requests use different rates." },
@@ -411,15 +438,18 @@ var defaultPricing = {
   "claude-opus-4-5": { inputPerMillion: 5, cacheCreationInput5mPerMillion: 6.25, cacheCreationInput1hPerMillion: 10, cacheCreationInputPerMillion: 10, cachedInputPerMillion: 0.5, outputPerMillion: 25 },
   "claude-opus-4-1": { inputPerMillion: 15, cacheCreationInput5mPerMillion: 18.75, cacheCreationInput1hPerMillion: 30, cacheCreationInputPerMillion: 30, cachedInputPerMillion: 1.5, outputPerMillion: 75 },
   "claude-opus-4": { inputPerMillion: 15, cacheCreationInput5mPerMillion: 18.75, cacheCreationInput1hPerMillion: 30, cacheCreationInputPerMillion: 30, cachedInputPerMillion: 1.5, outputPerMillion: 75 },
-  "claude-sonnet-5-5": { inputPerMillion: 2, cacheCreationInput5mPerMillion: 2.5, cacheCreationInput1hPerMillion: 4, cacheCreationInputPerMillion: 4, cachedInputPerMillion: 0.2, outputPerMillion: 10 },
+  "claude-sonnet-5-5": { inputPerMillion: 2, cacheCreationInput5mPerMillion: 2.5, cacheCreationInput1hPerMillion: 4, cacheCreationInputPerMillion: 4, cachedInputPerMillion: 0.1, outputPerMillion: 10 },
   "claude-sonnet-5": { inputPerMillion: 2, cacheCreationInput5mPerMillion: 2.5, cacheCreationInput1hPerMillion: 4, cacheCreationInputPerMillion: 4, cachedInputPerMillion: 0.2, outputPerMillion: 10, note: "Anthropic made the $2 input / $10 output introductory pricing permanent on August 10, 2026." },
   "claude-sonnet-4-6": { inputPerMillion: 3, cacheCreationInput5mPerMillion: 3.75, cacheCreationInput1hPerMillion: 6, cacheCreationInputPerMillion: 6, cachedInputPerMillion: 0.3, outputPerMillion: 15 },
   "claude-sonnet-4-5": { inputPerMillion: 3, cacheCreationInput5mPerMillion: 3.75, cacheCreationInput1hPerMillion: 6, cacheCreationInputPerMillion: 6, cachedInputPerMillion: 0.3, outputPerMillion: 15 },
   "claude-sonnet-4": { inputPerMillion: 3, cacheCreationInput5mPerMillion: 3.75, cacheCreationInput1hPerMillion: 6, cacheCreationInputPerMillion: 6, cachedInputPerMillion: 0.3, outputPerMillion: 15 },
   "claude-haiku-4-5": { inputPerMillion: 1, cacheCreationInput5mPerMillion: 1.25, cacheCreationInput1hPerMillion: 2, cacheCreationInputPerMillion: 2, cachedInputPerMillion: 0.1, outputPerMillion: 5 },
+  "claude-haiku-5-5": { inputPerMillion: 0.1, cacheCreationInput5mPerMillion: 0.125, cacheCreationInput1hPerMillion: 0.2, cacheCreationInputPerMillion: 0.2, cachedInputPerMillion: 0.01, outputPerMillion: 0.5, note: "Prompts up to 100,000 input tokens per request, including cache reads and writes. Larger prompts use the separately editable claude-haiku-5-5-long-context card. Aggregate-only usage stays unpriced when request lengths are unknown." },
+  "claude-haiku-5-5-long-context": { inputPerMillion: 0.5, cacheCreationInput5mPerMillion: 0.625, cacheCreationInput1hPerMillion: 1, cacheCreationInputPerMillion: 1, cachedInputPerMillion: 0.05, outputPerMillion: 2.5, note: "RepoSpend pricing card, not a provider model ID. Applied to the whole Haiku 5.5 request when total prompt length exceeds 100,000 tokens." },
   "claude-3-5-haiku": { inputPerMillion: 0.8, cacheCreationInput5mPerMillion: 1, cacheCreationInput1hPerMillion: 1.6, cacheCreationInputPerMillion: 1.6, cachedInputPerMillion: 0.08, outputPerMillion: 4 }
 };
 var legacyBundledPricing = {
+  "claude-sonnet-5-5": { inputPerMillion: 2, cacheCreationInput5mPerMillion: 2.5, cacheCreationInput1hPerMillion: 4, cacheCreationInputPerMillion: 4, cachedInputPerMillion: 0.2, outputPerMillion: 10 },
   "gpt-5.6": { inputPerMillion: 5, cachedInputPerMillion: 0.5, cacheCreationInputPerMillion: 6.25, outputPerMillion: 30, reasoningOutputPerMillion: 30, note: "GPT-5.6 Sol flagship tier. OpenAI preview pricing lists Sol at $5 input / $30 output per 1M tokens, cache writes at 1.25x input, and cache reads at a 90% discount." },
   "gpt-5.6-sol": { inputPerMillion: 5, cachedInputPerMillion: 0.5, cacheCreationInputPerMillion: 6.25, outputPerMillion: 30, reasoningOutputPerMillion: 30 },
   "gpt-5.6-terra": { inputPerMillion: 2.5, cachedInputPerMillion: 0.25, cacheCreationInputPerMillion: 3.125, outputPerMillion: 15, reasoningOutputPerMillion: 15 },
@@ -526,12 +556,43 @@ function savePricingTable(pricingPath, pricing) {
 function repospendHome() {
   return process.env.REPOSPEND_HOME || path.join(os.homedir(), ".repospend");
 }
-function calculateCostUsd(usage, pricing) {
+function calculateCostUsd(usage, pricing, requestInputTokens) {
+  const cost = unroundedCostUsd(usage, pricing, requestInputTokens);
+  return cost === void 0 ? void 0 : Number((cost + Number.EPSILON).toFixed(6));
+}
+function calculateRequestCosts(requests, pricing) {
+  const resolutions = /* @__PURE__ */ new Map();
+  const warnings = /* @__PURE__ */ new Set();
+  let total = 0;
+  for (const request of requests) {
+    let resolution = request.model ? resolutions.get(request.model) : void 0;
+    if (request.model && !resolution) {
+      resolution = resolvePricingForModel(request.model, pricing);
+      resolutions.set(request.model, resolution);
+    }
+    const cost = unroundedCostUsd(request, pricing, request.requestInputTokens, resolution);
+    if (cost !== void 0) total += cost;
+    else if (request.model && resolution) {
+      const missingContext = needsHaikuContext(request.model, resolution) && (request.requestInputTokens === void 0 || !Number.isFinite(request.requestInputTokens) || request.requestInputTokens < 0);
+      warnings.add(missingContext ? "unknown_request_context" : `unknown_pricing_model:${normalizePricingModelId(request.model)}`);
+    } else warnings.add("unknown_pricing");
+  }
+  return { costUsd: requests.length && !warnings.size ? Number((total + Number.EPSILON).toFixed(6)) : void 0, warnings: [...warnings] };
+}
+function needsHaikuContext(model, resolution) {
+  return resolution.sourceModel === "claude-haiku-5-5" || resolution.inherited && /^claude-haiku-5-5(?:$|[-.@[])/.test(normalizePricingModelId(model));
+}
+function unroundedCostUsd(usage, pricing, requestInputTokens, cachedResolution) {
   if (!usage.model) {
     return void 0;
   }
-  const modelPricing = findModelPricing(usage.model, pricing);
-  if (!modelPricing) {
+  const resolution = cachedResolution ?? resolvePricingForModel(usage.model, pricing);
+  let modelPricing = resolution.pricing;
+  if (needsHaikuContext(usage.model, resolution)) {
+    if (requestInputTokens === void 0 || !Number.isFinite(requestInputTokens) || requestInputTokens < 0) return void 0;
+    modelPricing = pricing[requestInputTokens > 1e5 ? "claude-haiku-5-5-long-context" : "claude-haiku-5-5"];
+  }
+  if (!isUsableModelPricing(modelPricing)) {
     return void 0;
   }
   const cacheCreationInput5m = usage.cacheCreationInputTokens5m ?? 0;
@@ -546,7 +607,7 @@ function calculateCostUsd(usage, pricing) {
   const cacheCreation1hRate = firstPositiveRate(modelPricing.cacheCreationInput1hPerMillion, modelPricing.cacheCreationInputPerMillion, modelPricing.inputPerMillion);
   const reasoningRate = modelPricing.reasoningOutputPerMillion ?? modelPricing.outputPerMillion;
   const cost = billableInput / 1e6 * modelPricing.inputPerMillion + usage.cachedInputTokens / 1e6 * cachedRate + cacheCreationInput5m / 1e6 * cacheCreation5mRate + cacheCreationInput1h / 1e6 * cacheCreation1hRate + unclassifiedCacheCreationInput / 1e6 * cacheCreationRate + usage.outputTokens / 1e6 * modelPricing.outputPerMillion + usage.reasoningTokens / 1e6 * reasoningRate;
-  return Number(cost.toFixed(6));
+  return cost;
 }
 function firstPositiveRate(...rates) {
   return rates.find((rate) => typeof rate === "number" && Number.isFinite(rate) && rate > 0) ?? 0;
@@ -2285,12 +2346,13 @@ function threadToUsage(row, index, sessionFiles, options, importedModels = /* @_
     sessionOutcome: inferOutcome(sessionBreakdown),
     compaction: sessionBreakdown.compaction
   };
-  const cost = hasPricedBreakdown ? calculateCostUsd(usage, options.pricing) : void 0;
+  const estimate = hasPricedBreakdown ? calculateRequestCosts([usage], options.pricing) : void 0;
+  const cost = estimate?.costUsd;
   const costWarnings = hasPricedBreakdown ? warnings : [...warnings, "missing_token_breakdown"];
   return {
     ...usage,
     estimatedCostUsd: cost,
-    warnings: cost === void 0 ? [...costWarnings, "unknown_pricing"] : warnings
+    warnings: cost === void 0 ? [.../* @__PURE__ */ new Set([...costWarnings, "unknown_pricing", ...estimate?.warnings ?? []])] : warnings
   };
 }
 function readStandaloneSessions(sessionFiles, options) {
@@ -2360,11 +2422,12 @@ function readStandaloneSessions(sessionFiles, options) {
       compaction: breakdown.compaction
     };
     const hasPricedBreakdown = breakdown.inputTokens + breakdown.cachedInputTokens + breakdown.outputTokens + breakdown.reasoningTokens > 0;
-    const cost = hasPricedBreakdown ? calculateCostUsd(usage, options.pricing) : void 0;
+    const estimate = hasPricedBreakdown ? calculateRequestCosts([usage], options.pricing) : void 0;
+    const cost = estimate?.costUsd;
     return {
       ...usage,
       estimatedCostUsd: cost,
-      warnings: cost === void 0 ? [...usage.warnings, hasPricedBreakdown ? "unknown_pricing" : "missing_token_breakdown"] : usage.warnings
+      warnings: cost === void 0 ? [.../* @__PURE__ */ new Set([...usage.warnings, hasPricedBreakdown ? "unknown_pricing" : "missing_token_breakdown", ...estimate?.warnings ?? []])] : usage.warnings
     };
   });
 }
@@ -3050,21 +3113,30 @@ function claudeBreakdownToUsage(file, index, options, breakdown, idSuffix = "")
     promptTimeline: breakdown.promptTimeline,
     sessionOutcome: inferOutcome2(breakdown)
   };
-  const costBuckets = /* @__PURE__ */ new Map();
-  for (const event of breakdown.usageEvents) {
+  const requests = breakdown.usageEvents.map((event) => {
     const model = event.model ?? breakdown.model;
     const normalized = model ? normalizePricingModelId(model) : void 0;
     const pricingModel = normalized && normalizedServiceTier(event.usage.speed) === "fast" && !normalized.includes("fast-mode") && !/^claude-opus-4-6(?:$|-)/.test(normalized) ? `${normalized}-fast-mode` : model;
-    const totals = costBuckets.get(pricingModel) ?? emptyBreakdown2();
+    const totals = emptyBreakdown2();
     applyUsage(totals, event.usage);
-    costBuckets.set(pricingModel, totals);
-  }
-  const costs = hasTokenBreakdown ? [...costBuckets].map(([model, totals]) => calculateCostUsd({ ...totals, model, reasoningTokens: 0 }, options.pricing)) : [];
-  const cost = costs.length && costs.every((value) => value !== void 0) ? Number(costs.reduce((sum2, value) => sum2 + (value ?? 0), 0).toFixed(6)) : void 0;
+    return {
+      model: pricingModel,
+      inputTokens: totals.inputTokens,
+      cachedInputTokens: totals.cachedInputTokens,
+      cacheCreationInputTokens: totals.cacheCreationInputTokens,
+      cacheCreationInputTokens5m: totals.cacheCreationInputTokens5m,
+      cacheCreationInputTokens1h: totals.cacheCreationInputTokens1h,
+      outputTokens: totals.outputTokens,
+      reasoningTokens: 0,
+      requestInputTokens: hasInputTokenCount(event.usage) ? totals.inputTokens : void 0
+    };
+  });
+  const estimate = hasTokenBreakdown ? calculateRequestCosts(requests, options.pricing) : void 0;
+  const cost = estimate?.costUsd;
   return {
     ...usage,
     estimatedCostUsd: cost,
-    warnings: cost === void 0 ? [...usage.warnings, "unknown_pricing", ...costs.some((value, index2) => value === void 0 && [...costBuckets.keys()][index2]?.includes("fast-mode")) ? ["unknown_fast_mode_pricing"] : []] : usage.warnings
+    warnings: cost === void 0 ? [.../* @__PURE__ */ new Set([...usage.warnings, "unknown_pricing", ...estimate?.warnings ?? [], ...estimate?.warnings.some((warning) => warning.startsWith("unknown_pricing_model:") && warning.includes("fast-mode")) ? ["unknown_fast_mode_pricing"] : []])] : usage.warnings
   };
 }
 function inferClaudeEntrypointFromPath(filePath) {
@@ -3166,10 +3238,14 @@ function usageDedupKey(record, message) {
 function occurrenceKey(filePath, recordIndex) {
   return `${path6.resolve(filePath)}:${recordIndex}`;
 }
+function hasInputTokenCount(usage) {
+  const value = usage.input_tokens;
+  return (typeof value === "number" || typeof value === "string" && value.trim() !== "") && Number.isFinite(Number(value)) && Number(value) >= 0;
+}
 function cloneUsage(usage) {
   const cacheCreation = firstObject4(usage.cache_creation);
   return {
-    input_tokens: numberValue(usage.input_tokens),
+    input_tokens: hasInputTokenCount(usage) ? numberValue(usage.input_tokens) : void 0,
     cache_read_input_tokens: numberValue(usage.cache_read_input_tokens),
     cache_creation_input_tokens: numberValue(usage.cache_creation_input_tokens),
     cache_creation: cacheCreation ? {
@@ -3184,7 +3260,7 @@ function cloneUsage(usage) {
 function mergeUsageMax(target, usage) {
   const targetCacheCreation = firstObject4(target.cache_creation) ?? {};
   const usageCacheCreation = firstObject4(usage.cache_creation) ?? {};
-  target.input_tokens = Math.max(numberValue(target.input_tokens), numberValue(usage.input_tokens));
+  if (hasInputTokenCount(target) || hasInputTokenCount(usage)) target.input_tokens = Math.max(numberValue(target.input_tokens), numberValue(usage.input_tokens));
   target.cache_read_input_tokens = Math.max(numberValue(target.cache_read_input_tokens), numberValue(usage.cache_read_input_tokens));
   target.cache_creation_input_tokens = Math.max(numberValue(target.cache_creation_input_tokens), numberValue(usage.cache_creation_input_tokens));
   target.cache_creation = {
@@ -3767,6 +3843,8 @@ function usageCandidateFromRecord(record, index, fallbackTimestamp, traceContext
   const output = numberValue2(attributes["gen_ai.usage.output_tokens"]);
   const cached = numberValue2(attributes["gen_ai.usage.cache_read.input_tokens"]);
   const cacheCreation = firstPositiveNumber(attributes, ["gen_ai.usage.cache_write.input_tokens", "gen_ai.usage.cache_creation.input_tokens"]);
+  const rawInput = attributes["gen_ai.usage.input_tokens"];
+  const requestInputTokens = typeof rawInput === "number" && Number.isFinite(rawInput) && rawInput >= cached + cacheCreation ? rawInput : void 0;
   const reasoning = firstPositiveNumber(attributes, ["gen_ai.usage.reasoning.output_tokens", "gen_ai.usage.reasoning_tokens"]);
   const total = firstPositiveNumber(attributes, ["gen_ai.usage.total_tokens", "gen_ai.usage.total.token_count"]);
   const outputWithFallback = output || outputTokensFromTotal(total, input, reasoning);
@@ -3783,6 +3861,7 @@ function usageCandidateFromRecord(record, index, fallbackTimestamp, traceContext
     model: firstNonEmptyAttr(attributes, modelAttrs) ?? traceContext?.model,
     timestamp: timestamp3,
     durationMs: durationMsFromRecord(record),
+    requestInputTokens,
     inputTokens: input,
     cachedInputTokens: cached,
     cacheCreationInputTokens: cacheCreation,
@@ -3845,6 +3924,8 @@ function applyCandidate(breakdown, candidate) {
   breakdown.endedAt = maxDate2(breakdown.endedAt, candidate.timestamp);
   applyTokenEvent(breakdown, {
     dedupKey: candidate.dedupKey,
+    model: candidate.model,
+    requestInputTokens: candidate.source === "chat_span" || candidate.source === "inference_log" ? candidate.requestInputTokens : void 0,
     detailed: candidate.inputTokens > 0 || candidate.cachedInputTokens > 0 || candidate.cacheCreationInputTokens > 0 || candidate.reasoningTokens > 0,
     inputTokens: candidate.inputTokens,
     cachedInputTokens: candidate.cachedInputTokens,
@@ -3966,8 +4047,9 @@ function breakdownToUsage(breakdown, index, options) {
     fileEditCount: breakdown.fileEditCount,
     sourceMetadata: { copilot: { sourcePaths: breakdown.sourcePaths, surface: breakdown.surface } }
   };
-  const cost = hasPricedBreakdown ? calculateCostUsd(usage, options.pricing) : void 0;
-  return { ...usage, estimatedCostUsd: cost, warnings: hasPricedBreakdown && cost === void 0 ? [...usage.warnings, "unknown_pricing"] : usage.warnings };
+  const estimate = hasPricedBreakdown ? calculateRequestCosts(breakdown.tokenEvents.map((event) => ({ ...event, model: normalizeCopilotModel(hasKnownModel(event.model) ? event.model : breakdown.model) })), options.pricing) : void 0;
+  const cost = estimate?.costUsd;
+  return { ...usage, estimatedCostUsd: cost, warnings: hasPricedBreakdown && cost === void 0 ? [.../* @__PURE__ */ new Set([...usage.warnings, "unknown_pricing", ...estimate?.warnings ?? []])] : usage.warnings };
 }
 function unknownCopilotRepoInfo() {
   return {
@@ -4872,12 +4954,13 @@ function candidateToUsage(candidate, index, options) {
       }
     }
   };
-  const pricedCost = hasTokenBreakdown ? calculateCostUsd(usage, options.pricing) : void 0;
+  const estimate = hasTokenBreakdown ? calculateRequestCosts([usage], options.pricing) : void 0;
+  const pricedCost = estimate?.costUsd;
   const estimatedCostUsd = pricedCost ?? breakdown.estimatedCostUsd;
   return {
     ...usage,
     estimatedCostUsd,
-    warnings: estimatedCostUsd === void 0 ? [...usage.warnings, breakdown.hasEstimatedPromptTokens ? "cursor_cost_not_estimated_from_prompt_tokens" : hasTokenBreakdown ? "unknown_pricing" : "unknown_cost"] : usage.warnings
+    warnings: estimatedCostUsd === void 0 ? [.../* @__PURE__ */ new Set([...usage.warnings, breakdown.hasEstimatedPromptTokens ? "cursor_cost_not_estimated_from_prompt_tokens" : hasTokenBreakdown ? "unknown_pricing" : "unknown_cost", ...estimate?.warnings ?? []])] : usage.warnings
   };
 }
 function inferProjectPathFromCursorPath(filePath, cursorHome) {

package/docs/llms.txt

@@ -9,6 +9,8 @@ Canonical URLs:
 - npm: https://www.npmjs.com/package/repospend
 - Package name: repospend
 - Run command: npx repospend
+- GPT-6.1 Sol pricing and Codex tracking: https://repospend.com/guides/gpt-6-1-sol-pricing
+- AI coding price comparison: https://repospend.com/guides/ai-coding-price-war
 
 What RepoSpend is:
 - A local-first AI coding token usage dashboard.
@@ -56,5 +58,10 @@ Positioning:
 - RepoSpend totals can intentionally differ from ccusage or Tokscale because it includes Claude Desktop/local-agent session roots when present, treats cache reads/writes as input sub-buckets, prices Claude cache writes by recorded 5-minute vs 1-hour TTL, and separates Codex visible output from reasoning output.
 
 Cost language:
+- October 17, 2026 pricing update: Sonnet 5.5 cache reads are $0.10 per million tokens; Sonnet 5 remains at $0.20.
+- Haiku 5.5 has two per-request prompt-length tiers: input/cache read/output $0.10/$0.01/$0.50 up to 100K, and $0.50/$0.05/$2.50 above 100K. Cache writes have separate 5-minute and 1-hour rates. Unknown request lengths stay unpriced.
+- Gemini 4 Argon introductory input/cache read/output rates are $2/$0.10/$10 through November 30, 2026, then published rates become $4/$0.20/$20. Bundled tables are a dated snapshot, not an automatic future rate switch.
+- Current API-equivalent pricing applied to older usage is not historical billing reconstruction.
+- GPT-6.1 Sol has an explicit Standard short-context card: $2 input, $0.10 cached input, $2.50 cache writes, $10 output per million tokens, checked September 29, 2026.
 - RepoSpend shows estimated API-equivalent cost.
 - RepoSpend does not claim to show actual bills, invoices, subscription usage, savings, credits, or account-specific charges.

package/docs/pricing.md

@@ -4,10 +4,35 @@ RepoSpend estimates API-equivalent cost from a local pricing table. The bundled
 
 The bundled table is seeded from public OpenAI, Anthropic, Google, and GitHub Copilot model references and is expressed as USD per 1M tokens. Pricing changes over time, so treat RepoSpend costs as API-equivalent estimates rather than invoice-grade accounting.
 
-The bundled defaults were checked on September 28, 2026 and use published Standard API prices. Claude Sonnet 5.5 has explicit $2 input / $10 output rates with the same cache rates as Sonnet 5. Claude Sonnet 5 remains at $2 input and $10 output per 1M tokens because Anthropic made its introductory rate permanent on August 10, 2026.
+The October 17, 2026 update checks Claude Haiku 5.5, Sonnet 5.5, and Gemini 4 Argon against provider documentation. The GPT-6.1 Sol card was checked on September 29; the other bundled defaults retain the September 28 snapshot. Defaults use published Standard API prices. Sonnet 5.5 cache reads fell from $0.20 to $0.10 per 1M tokens on October 7; its $2 input, $10 output, $2.50 5-minute writes, and $4 1-hour writes are unchanged. Sonnet 5 retains its $0.20 cache-read rate.
+
+### Claude Haiku 5.5 and request-length pricing
+
+All rates are USD per 1M tokens, using Anthropic's global Standard API prices:
+
+| Prompt length per request | Input | Cache reads | 5m writes | 1h writes | Output |
+| --- | ---: | ---: | ---: | ---: | ---: |
+| Up to and including 100,000 tokens | $0.10 | $0.01 | $0.125 | $0.20 | $0.50 |
+| Over 100,000 tokens | $0.50 | $0.05 | $0.625 | $1 | $2.50 |
+
+Prompt length includes uncached input, cache reads, and cache writes. The higher rates apply to the whole qualifying request, not just tokens above 100K. Multiple short requests keep the lower rates even when combined session input exceeds 100K. Claude's deduplicated usage events and Copilot chat/inference records are priced individually, then their unrounded costs are summed before rounding the session estimate once.
+
+Settings exposes `claude-haiku-5-5` for the lower tier and `claude-haiku-5-5-long-context` for the higher tier. The latter is a RepoSpend rate card, not a provider model ID; both can be edited separately. The higher-tier card appears with used Haiku models. Editing an inherited Haiku snapshot updates the shared base card and retains automatic tier selection. Use Add model to intentionally create an exact custom snapshot with flat rates; such overrides remain authoritative. Haiku usage without a known request length (including aggregate-only Copilot, Codex, or experimental Cursor totals) stays unpriced unless an explicit custom flat snapshot card or recorded Cursor cost is available. Pricing triage distinguishes missing request lengths from missing rates and exposes any unpriced auxiliary request models. Unknown fast-mode cards remain unpriced.
+
+These are current API-equivalent rates applied to the selected usage window, not historical date-effective billing. The Sonnet reduction does not claim that pre-October 7 API requests were billed at the new price. Sources: [Anthropic pricing](https://platform.claude.com/docs/en/about-claude/pricing), [October 7 announcement](https://www.anthropic.com/claude-haiku-5-5).
+
+### Gemini 4 Argon introductory rates
+
+`gemini-4-argon`, including Copilot-prefixed/suffixed aliases, uses $2 input, $0.10 cached input, and $10 output (including thinking) per 1M tokens through November 30, 2026. Google lists $4 input, $0.20 cached input, and $20 output from December 1. Argon has no explicit cache-write charge. Bundled rates are a verified snapshot: the application does not silently switch prices based on the machine clock. Refresh the table at the promotion boundary or save intentional overrides. Source: [Google's model and pricing guide](https://ai.google.dev/gemini-api/docs/latest-model?hl=en).
 
 GPT-6 Astra, Sol, and Luna use OpenAI's Standard short-context rates. Their long-context requests above 272,000 input tokens use different rates, and local session totals do not identify every request's pricing tier. Fast, batch, and regional processing can also differ from these defaults. RepoSpend keeps the result labeled as an API-equivalent estimate.
 
+### GPT-6.1 Sol (September 29, 2026)
+
+The explicit `gpt-6.1-sol` card uses $2 input, $0.10 cached input, $2.50 cache writes, and $10 output/reasoning per 1M tokens, verified against [OpenAI's model reference](https://developers.openai.com/api/docs/models/gpt-6.1-sol) on September 29. GPT-6 Sol remains separate at $0.20 for cached input. Case-insensitive and Copilot-prefixed IDs resolve to the same card; terminal snapshot dates preserve the model generation. Snapshot support is defensive parsing, not a claim that OpenAI publishes those snapshot IDs. Custom exact-ID overrides continue to take precedence.
+
+Above 272K input tokens per request, GPT-6.1 Sol's Standard rates become $4 input, $0.20 cached input, $5 cache writes, and $15 output per 1M tokens. Fast costs twice Standard; Batch and Flex cost half Standard. RepoSpend's default estimate uses Standard short-context pricing because aggregate local usage cannot reliably identify these modifiers for every request. See the [GPT-6.1 Sol pricing and Codex tracking guide](https://repospend.com/guides/gpt-6-1-sol-pricing) for a worked example.
+
 Claude Fable 5.1 and Mythos 5.1 retain their predecessor's base rates but have lower cache-read pricing. Claude Opus 5.5 has its own lower base and cache rates. These model rows use Anthropic's global Standard rates.
 
 OpenAI reduced GPT-5.6 pricing effective July 30, 2026. RepoSpend uses $2 input and $12 output per 1M tokens for Terra, and $0.20 input and $1.20 output for Luna. Cached input remains 90% below the uncached input rate, and cache writes are billed at 1.25x the uncached input rate, so Luna's bundled cache-write rate is $0.25 per 1M tokens.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "repospend",
-  "version": "0.1.5",
+  "version": "0.1.7",
   "description": "Local-first dashboard for tracking AI coding token usage and API-equivalent spend by repository, session, model, and tool.",
   "license": "Apache-2.0",
   "author": "Mehmet Mustafa Demir",