JCV's Portfolio

src/llm/pricing.ts

// Pricing tables are manually maintained in this file. // Last updated: 2026-03-18. const DEFAULT_PRICING = { openai: { "gpt-5.5": { inputPer1M: 5.0, cacheReadPer1M: 0.5, outputPer1M: 30.0 }, "gpt-5.2": { inputPer1M: 1.75, cacheReadPer1M: 0.175, outputPer1M: 14.0 }, "gpt-5.1": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-5.4": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-5.4-mini": { inputPer1M: 0.75, cacheReadPer1M: 0.075, outputPer1M: 4.5 }, "gpt-5.4-nano": { inputPer1M: 0.2, cacheReadPer1M: 0.02, outputPer1M: 1.25 }, "gpt-5": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-5-mini": { inputPer1M: 0.25, cacheReadPer1M: 0.025, outputPer1M: 2.0 }, "gpt-5-nano": { inputPer1M: 0.05, cacheReadPer1M: 0.005, outputPer1M: 0.4 }, "gpt-5.2-chat-latest": { inputPer1M: 1.75, cacheReadPer1M: 0.175, outputPer1M: 14.0 }, "gpt-5.1-chat-latest": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-5-chat-latest": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-5.3-codex": { inputPer1M: 1.75, cacheReadPer1M: 0.175, outputPer1M: 14.0 }, "gpt-5.2-codex": { inputPer1M: 1.75, cacheReadPer1M: 0.175, outputPer1M: 14.0 }, "gpt-5.1-codex-max": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-5.1-codex": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-5-codex": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-5.1-codex-mini": { inputPer1M: 0.25, cacheReadPer1M: 0.025, outputPer1M: 2.0 }, "codex-mini-latest": { inputPer1M: 1.5, cacheReadPer1M: 0.375, outputPer1M: 6.0 }, "gpt-5.2-pro": { inputPer1M: 21.0, outputPer1M: 168.0 }, "gpt-5-pro": { inputPer1M: 15.0, outputPer1M: 120.0 }, "gpt-realtime": { inputPer1M: 4.0, cacheReadPer1M: 0.4, outputPer1M: 16.0 }, "gpt-realtime-1.5": { inputPer1M: 4.0, cacheReadPer1M: 0.4, outputPer1M: 16.0 }, "gpt-realtime-mini": { inputPer1M: 0.6, cacheReadPer1M: 0.06, outputPer1M: 2.4 }, "gpt-4o-realtime-preview": { inputPer1M: 5.0, cacheReadPer1M: 2.5, outputPer1M: 20.0 }, "gpt-4o-mini-realtime-preview": { inputPer1M: 0.6, cacheReadPer1M: 0.3, outputPer1M: 2.4 }, "gpt-audio": { inputPer1M: 2.5, outputPer1M: 10.0 }, "gpt-audio-1.5": { inputPer1M: 2.5, outputPer1M: 10.0 }, "gpt-audio-mini": { inputPer1M: 0.6, outputPer1M: 2.4 }, "gpt-4o-audio-preview": { inputPer1M: 2.5, outputPer1M: 10.0 }, "gpt-4o-mini-audio-preview": { inputPer1M: 0.15, outputPer1M: 0.6 }, "o4-mini": { inputPer1M: 1.1, cacheReadPer1M: 0.275, outputPer1M: 4.4 }, "o4-mini-deep-research": { inputPer1M: 2.0, cacheReadPer1M: 0.5, outputPer1M: 8.0 }, "gpt-5-search-api": { inputPer1M: 1.25, cacheReadPer1M: 0.125, outputPer1M: 10.0 }, "gpt-4o-mini-search-preview": { inputPer1M: 0.15, outputPer1M: 0.6 }, "gpt-4o-search-preview": { inputPer1M: 2.5, outputPer1M: 10.0 }, "text-embedding-3-small": { inputPer1M: 0.02, outputPer1M: 0 }, "text-embedding-3-large": { inputPer1M: 0.13, outputPer1M: 0 }, "text-embedding-ada-002": { inputPer1M: 0.1, outputPer1M: 0 } }, openaiImages: { "gpt-image-2": { "1024x1024": 0.053, "1024x1536": 0.041, "1536x1024": 0.041 }, "gpt-image-1.5": { "1024x1024": 0.04 } }, anthropic: { "claude-opus-4-6": { inputPer1M: 5.0, cacheWritePer1M: 6.25, cacheWrite1hPer1M: 10.0, cacheReadPer1M: 0.5, outputPer1M: 25.0 }, "claude-opus-4-5": { inputPer1M: 5.0, cacheWritePer1M: 6.25, cacheWrite1hPer1M: 10.0, cacheReadPer1M: 0.5, outputPer1M: 25.0 }, "claude-opus-4-1": { inputPer1M: 15.0, cacheWritePer1M: 18.75, cacheWrite1hPer1M: 30.0, cacheReadPer1M: 1.5, outputPer1M: 75.0 }, "claude-opus-4": { inputPer1M: 15.0, cacheWritePer1M: 18.75, cacheWrite1hPer1M: 30.0, cacheReadPer1M: 1.5, outputPer1M: 75.0 }, "claude-sonnet-4-6": { inputPer1M: 3.0, cacheWritePer1M: 3.75, cacheWrite1hPer1M: 6.0, cacheReadPer1M: 0.3, outputPer1M: 15.0 }, "claude-sonnet-4-5": { inputPer1M: 3.0, cacheWritePer1M: 3.75, cacheWrite1hPer1M: 6.0, cacheReadPer1M: 0.3, outputPer1M: 15.0 }, "claude-sonnet-4": { inputPer1M: 3.0, cacheWritePer1M: 3.75, cacheWrite1hPer1M: 6.0, cacheReadPer1M: 0.3, outputPer1M: 15.0 }, "claude-sonnet-4-0": { inputPer1M: 3.0, cacheWritePer1M: 3.75, cacheWrite1hPer1M: 6.0, cacheReadPer1M: 0.3, outputPer1M: 15.0 }, "claude-3-7-sonnet-latest": { inputPer1M: 3.0, cacheWritePer1M: 3.75, cacheWrite1hPer1M: 6.0, cacheReadPer1M: 0.3, outputPer1M: 15.0 }, "claude-haiku-4-5": { inputPer1M: 1.0, cacheWritePer1M: 1.25, cacheWrite1hPer1M: 2.0, cacheReadPer1M: 0.1, outputPer1M: 5.0 }, "claude-3-5-sonnet-latest": { inputPer1M: 3.0, cacheWritePer1M: 3.75, cacheWrite1hPer1M: 6.0, cacheReadPer1M: 0.3, outputPer1M: 15.0 }, "claude-opus-3": { inputPer1M: 15.0, cacheWritePer1M: 18.75, cacheWrite1hPer1M: 30.0, cacheReadPer1M: 1.5, outputPer1M: 75.0 }, "claude-haiku-3": { inputPer1M: 0.25, cacheWritePer1M: 0.3, cacheWrite1hPer1M: 0.5, cacheReadPer1M: 0.03, outputPer1M: 1.25 } }, xai: { "grok-4-latest": { inputPer1M: 3.0, outputPer1M: 15.0 }, "grok-4-0709": { inputPer1M: 3.0, outputPer1M: 15.0 }, "grok-4": { inputPer1M: 3.0, outputPer1M: 15.0 }, "grok-4-fast-reasoning-latest": { inputPer1M: 0.2, outputPer1M: 0.5, cacheWritePer1M: 0.05 }, "grok-4-fast-reasoning": { inputPer1M: 0.2, outputPer1M: 0.5, cacheWritePer1M: 0.05 }, "grok-4-fast-reasoning:free": { inputPer1M: 0, outputPer1M: 0 }, "grok-code-fast-1-latest": { inputPer1M: 0.2, outputPer1M: 1.5, cacheWritePer1M: 0.05 }, "grok-code-fast-1": { inputPer1M: 0.2, outputPer1M: 1.5, cacheWritePer1M: 0.05 }, "grok-code-fast-1:free": { inputPer1M: 0, outputPer1M: 0 }, "grok-3-fast-latest": { inputPer1M: 5.0, outputPer1M: 25.0 }, "grok-3-fast": { inputPer1M: 5.0, outputPer1M: 25.0 }, "grok-3-fast-beta": { inputPer1M: 5.0, outputPer1M: 25.0 }, "grok-3-latest": { inputPer1M: 3.0, outputPer1M: 15.0 }, "grok-3": { inputPer1M: 3.0, outputPer1M: 15.0 }, "grok-3-beta": { inputPer1M: 3.0, outputPer1M: 15.0 }, "grok-3-mini-latest": { inputPer1M: 0.3, outputPer1M: 0.5 }, "grok-3-mini": { inputPer1M: 0.3, outputPer1M: 0.5 }, "grok-3-mini-beta": { inputPer1M: 0.3, outputPer1M: 0.5 }, "grok-3-mini-fast-latest": { inputPer1M: 0.3, outputPer1M: 0.5 }, "grok-3-mini-fast": { inputPer1M: 0.3, outputPer1M: 0.5 }, "grok-3-mini-fast-beta": { inputPer1M: 0.3, outputPer1M: 0.5 }, "grok-2-latest": { inputPer1M: 2.0, outputPer1M: 10.0 }, "grok-2-1212": { inputPer1M: 2.0, outputPer1M: 10.0 }, "grok-2": { inputPer1M: 2.0, outputPer1M: 10.0 }, "grok-2-vision-latest": { inputPer1M: 2.0, outputPer1M: 10.0 }, "grok-2-vision-1212": { inputPer1M: 2.0, outputPer1M: 10.0 }, "grok-2-vision": { inputPer1M: 2.0, outputPer1M: 10.0 }, "grok-beta": { inputPer1M: 5.0, outputPer1M: 15.0 }, "grok-vision-beta": { inputPer1M: 5.0, outputPer1M: 15.0 } }, "claude-oauth": { "claude-sonnet-4-6": { inputPer1M: 0, outputPer1M: 0 }, "claude-sonnet-4-5": { inputPer1M: 0, outputPer1M: 0 }, "claude-haiku-4-5": { inputPer1M: 0, outputPer1M: 0 }, "claude-opus-4-6": { inputPer1M: 0, outputPer1M: 0 }, "claude-opus-4-5": { inputPer1M: 0, outputPer1M: 0 } }, "openai-oauth": { "gpt-5.5": { inputPer1M: 0, outputPer1M: 0 }, "gpt-5.4": { inputPer1M: 0, outputPer1M: 0 }, "gpt-5.4-mini": { inputPer1M: 0, outputPer1M: 0 }, "gpt-5.4-nano": { inputPer1M: 0, outputPer1M: 0 }, "gpt-5.3-codex": { inputPer1M: 0, outputPer1M: 0 }, "gpt-5.1-codex-mini": { inputPer1M: 0, outputPer1M: 0 } }, "codex-cli": { "gpt-5.4": { inputPer1M: 0, outputPer1M: 0 }, "gpt-5.4-mini": { inputPer1M: 0, outputPer1M: 0 }, "gpt-5.4-nano": { inputPer1M: 0, outputPer1M: 0 }, "gpt-5-codex": { inputPer1M: 0, outputPer1M: 0 } }, xaiImages: { "grok-2-image-latest": { default: 0.07 }, "grok-2-image-1212": { default: 0.07 }, "grok-2-image": { default: 0.07 } } };

const MODEL_ALIASES = { "claude opus 4.6": "claude-opus-4-6", "claude opus 4.5": "claude-opus-4-5", "claude opus 4.1": "claude-opus-4-1", "claude opus 4": "claude-opus-4", "claude sonnet 4.6": "claude-sonnet-4-6", "claude sonnet 4.5": "claude-sonnet-4-5", "claude sonnet 4": "claude-sonnet-4", "claude sonnet 4.0": "claude-sonnet-4-0", "claude sonnet 3.7": "claude-3-7-sonnet-latest", "claude haiku 4.5": "claude-haiku-4-5", "claude opus 3": "claude-opus-3", "claude haiku 3": "claude-haiku-3", "grok 4": "grok-4-latest", "grok 4 fast reasoning": "grok-4-fast-reasoning-latest", "grok code fast 1": "grok-code-fast-1-latest", "grok 3": "grok-3-latest", "grok 3 fast": "grok-3-fast-latest", "grok 3 mini": "grok-3-mini-latest", "grok 3 mini fast": "grok-3-mini-fast-latest", "grok 2": "grok-2-latest", "grok 2 vision": "grok-2-vision-latest", "grok beta": "grok-beta", "gpt 5.4": "gpt-5.4", "gpt 5.4 mini": "gpt-5.4-mini", "gpt 5.4 nano": "gpt-5.4-nano", "grok vision beta": "grok-vision-beta" }; const LLM_PROVIDER_KEYS = ["openai", "anthropic", "xai", "claude-oauth", "openai-oauth", "codex-cli"]; const NON_TEXT_MODEL_PATTERNS = [ /embedding/i, /image/i, /video/i, /realtime/i, /audio/i, /(^|[-])search([:-]|$)/i ];

export function estimateUsdCost({ provider, model, inputTokens, outputTokens, cacheWriteTokens, cacheReadTokens, customPricing = {} }) { const resolvedProvider = normalizePricingProvider(provider); const merged = mergePricing(customPricing); const providerPricing = merged[resolvedProvider] ?? {}; const pricing = resolvePricing(providerPricing, model); if (!pricing) return 0;

const inputCost = toCost(inputTokens, pricing.inputPer1M); const outputCost = toCost(outputTokens, pricing.outputPer1M); const cacheWriteRate = Number(pricing.cacheWritePer1M ?? pricing.inputPer1M ?? 0); const cacheReadRate = Number(pricing.cacheReadPer1M ?? 0); const cacheWriteCost = toCost(cacheWriteTokens, cacheWriteRate); const cacheReadCost = toCost(cacheReadTokens, cacheReadRate); return Number((inputCost + outputCost + cacheWriteCost + cacheReadCost).toFixed(6)); }

export function estimateImageUsdCost({ provider, model, size = "1024x1024", imageCount = 1, customPricing = {} }) { const imageProviderKey = resolveImageProvider(provider); if (!imageProviderKey) return 0;

const merged = mergePricing(customPricing); const resolvedModel = resolveImageModelPricing(merged[imageProviderKey] ?? {}, model); if (!resolvedModel) return 0;

const normalizedSize = normalizeImageSize(size); const perImage = Number(resolvedModel[normalizedSize] ?? resolvedModel.default ?? 0); if (!perImage) return 0;

const count = Math.max(1, Math.floor(Number(imageCount) || 1)); return Number((perImage * count).toFixed(6)); }

export function getLlmModelCatalog(customPricing = {}) { const merged = mergePricing(customPricing); return { openai: listLlmModelsForProvider(merged.openai, "openai"), anthropic: listLlmModelsForProvider(merged.anthropic, "anthropic"), xai: listLlmModelsForProvider(merged.xai, "xai"), "claude-oauth": listLlmModelsForProvider(merged["claude-oauth"], "claude-oauth"), "openai-oauth": listLlmModelsForProvider(merged["openai-oauth"], "openai-oauth"), "codex-cli": listLlmModelsForProvider(merged["codex-cli"], "codex-cli") }; }

function normalizePricingProvider(provider) { const normalized = String(provider || "").trim().toLowerCase(); if (normalized === "codex-oauth") return "openai-oauth"; return normalized; }

function resolvePricing(providerPricing, model) { const exact = providerPricing[model]; if (exact) return exact;

const normalized = normalizeModelKey(model); if (!normalized) return null;

const alias = MODEL_ALIASES[normalized]; if (alias && providerPricing[alias]) return providerPricing[alias]; return providerPricing[normalized] ?? null; }

function resolveImageModelPricing(providerPricing, model) { const exact = providerPricing[model]; if (exact && typeof exact === "object") return exact;

const normalized = normalizeModelKey(model); if (!normalized) return null;

return providerPricing[normalized] ?? null; }

function normalizeModelKey(model) { return String(model || "") .trim() .toLowerCase() .replace(/^anthropic:/, "") .replace(/^xai:/, "") .replace(/\s*(deprecated)\s*/g, "") .replace(/\s+/g, " "); }

function resolveImageProvider(provider) { if (provider === "openai") return "openaiImages"; if (provider === "xai") return "xaiImages"; return null; }

function toCost(tokens, per1M) { return ((Number(tokens) || 0) / 1_000_000) * (Number(per1M) || 0); }

function normalizeImageSize(size) { return String(size || "") .trim() .toLowerCase() .replace(/\s+/g, ""); }

function listLlmModelsForProvider(providerPricing, provider) { if (!providerPricing || typeof providerPricing !== "object") return []; if (!LLM_PROVIDER_KEYS.includes(provider)) return [];

return Object.keys(providerPricing).filter((model) => isTextLlmModel(model, provider)); }

function isTextLlmModel(model, _provider) { const normalized = String(model || "").trim().toLowerCase(); if (!normalized) return false; if (NON_TEXT_MODEL_PATTERNS.some((pattern) => pattern.test(normalized))) return false; return true; }