AI Cost Calculators
Plug your usage into the right calculator and get the real monthly bill — not a vendor's marketing-friendly per-token number. Every page on this hub uses live 2026 prices pulled from the provider's pricing page, with batch (50% off) and cached-input (up to 90% off) discounts modelled out for you.
These are the single-topic pages ChatGPT and Perplexity cite when someone asks about a specific model's cost. Pick one to start, or sweep the grid to compare.
29 pages · updated 2026
Agent Loop Cost: Claude Opus 4.7 vs GPT-5 (2026) — Full $ Breakdown
Exact cost math for a 5-step agent loop on Claude Opus 4.7, Sonnet 4.6, GPT-5.5, and GPT-5.4. Formula, worked examples, cache impact, and multi-turn tables. Sourced June 2026.
ReadAI Incident Cost Calculator 2026: What an LLM Failure Actually Costs
Real-data calculator: what an AI incident costs in 2026. Direct response cost + customer churn + regulatory exposure (FTC, EU AI Act, state AGs) + brand impact. Sourced from the AI Incident Database (incidentdatabase.ai), publicized AI incidents 2023-2026, and applicable enforcement penalty schedules.
ReadAider Cost Per PR (2026): BYOK Math Across Claude, GPT-5, Gemini, DeepSeek
Aider is free open-source — but you pay model API. Real per-PR math across Claude Sonnet 4.6, Opus 4.7, GPT-5, GPT-5-mini, DeepSeek V3, and Gemini 2.5. Small fix ~$0.005, medium feature ~$0.08, large refactor ~$0.45 on Sonnet. Worked monthly spend at 1/5/20 PRs/day. When BYOK beats Cursor or Copilot subscription. Sourced from aider.chat.
ReadAlignment Tax Cost per Million Tokens (2026): The Real Number
The 'alignment tax' — extra tokens consumed by refusals, longer safety-tuned responses, instruction-hierarchy boilerplate, classifier pre-screening — quantified per model in 2026. Sourced from Anthropic, OpenAI, Google API list pricing and public model spec documentation.
ReadCursor vs Copilot Total Cost of Ownership (2026): Full TCO Math, 4 Worked Scenarios
The full TCO calculator for Cursor vs GitHub Copilot in 2026 — not just sticker price. Subscription + premium-credit overage + throttle-tax + onboarding + tool-sprawl. Worked dollar examples for solo dev, 5-person startup, 50-person org, and 500-person enterprise. Sourced from cursor.com/pricing and github.com/features/copilot/plans.
ReadDevin Cost Per Task (2026): ACU Math + Real-World $ Examples
Devin bills in Agent Compute Units (ACUs). Real per-task math for dependency upgrades, bug fixes, feature builds, and refactors. Pro $20 = ~10 ACUs, Max $200 = ~100 ACUs, Teams 100 + 40/user. When Devin's $/finished-task beats hiring a contractor — and when it doesn't. Sourced from devin.ai/pricing, June 2026.
ReadCohere vs OpenAI Embedding Cost (2026): embed-v4.0 vs text-embedding-3
Exact cost comparison of Cohere embed-v4.0 vs OpenAI text-embedding-3-small and text-embedding-3-large at 1M, 100M, and 1B tokens. Storage cost, dim tradeoffs, and when each wins. Sourced June 2026.
ReadFine-Tuning Cost by Model 2026: Real Per-Token Pricing Across 20+ Models
Fine-tuning cost breakdown by model in 2026: GPT-5, Claude Sonnet 4.6, Gemini 2.5 Flash, Llama 4 70B, Mistral, Qwen — per-1M-token training rates, inference markup, and deployment cost from live vendor docs.
ReadGDPR Compliance Cost for LLM Apps (2026): Full Budget Breakdown
What it actually costs to ship a GDPR-compliant LLM app in 2026 — legal, vendor, infra, ongoing. Itemized: DPO retainer or fractional, DPIA, vendor uplift (EU residency premium), SCCs/TIA legal review, sub-processor flow-down, audit + monitoring infra. Worked totals for solo founder, startup (Series A), and enterprise.
ReadLoRA Training Cost on H100 (2026): Real GPU-Hour Math for Llama 4, Mistral, Qwen
LoRA training cost on H100 GPUs in 2026: GPU-hour rates from Lambda, RunPod, CoreWeave, and per-job dollar cost for Llama 4 8B/70B/405B, Mistral, Qwen, and DeepSeek-V3 fine-tunes.
ReadMulti-Agent Cost Per Task (2026): Orchestrator + Workers Math
Exact cost math for multi-agent systems: orchestrator + 1/3/5 worker agents on CrewAI and LangGraph supervisor patterns. Reasoning model overhead included. Sourced June 2026.
ReadRAG Cost Per Query (2026): LLM + Vector DB + Embedding Full Breakdown
What does one RAG query actually cost? LLM call, vector search, query embedding, and optional reranking — all four layers broken down. Worked examples at 1K, 100K, and 1M queries/month. Sourced, June 2026.
ReadSOC 2 Prep Cost for AI Startups (2026): Type 1 + Type 2 Full Budget
What SOC 2 Type 1 and Type 2 prep actually costs an AI startup in 2026 — auditor fees, vendor compliance platform, security tooling, time-on-task, and the LLM-specific scope additions (prompt logging, vector DB controls, AI governance). Worked totals for seed, Series A, and Series B startups.
ReadTool Use Overhead Cost (2026): Function Call Tokens + Schema Cost
Quantify the token cost of tool calling vs raw chat: schema tokens, function-call overhead, tool-result tokens, parallel calls, and multi-turn loops. Real $ math, sourced June 2026.
ReadVector DB Cost per 1M Embeddings (2026): Pinecone, Weaviate, Qdrant, Turbopuffer
Exact cost to store and query 1M, 100M, and 1B embeddings across Pinecone, Weaviate, Qdrant, Zilliz, Turbopuffer, Chroma, and pgvector. Sourced from live vendor pricing pages, June 2026.
ReadClaude API Cost Calculator (2026): Opus 4.8, Sonnet 4.6, Haiku 4.5, Fable 5
Calculate exactly what a Claude API call costs in 2026. Live per-1M-token prices for Opus 4.8, Sonnet 4.6, Haiku 4.5, Fable 5. Prompt caching (5-min vs 1-hour TTL), Batch API (50% off), web search add-on. Worked $ examples and a sourced FAQ.
ReadCursor vs GitHub Copilot Cost (2026): Real $ Math, Every Plan, Credit Quotas
Cursor vs Copilot pricing head-to-head, June 2026. Cursor Pro $20, Business $40/seat. Copilot Pro $10, Pro+ $39, Max $100. Real $ math for solo devs, 5-person teams, and 50-person orgs. The credit-quota shift, the $30/mo Pro + Pro stack, and when Copilot Max beats Cursor Business.
ReadDeepSeek API Cost Calculator (2026): V3, R1, V4-Flash, V4-Pro Pricing
Calculate exactly what a DeepSeek API call costs in 2026. Live per-1M-token prices for DeepSeek-V3 ($0.14/$0.28), R1 ($0.55/$2.19), V4-Flash, V4-Pro. Cache-hit discounts (90%+ off). Worked $ examples vs OpenAI GPT-5.5 (35x cheaper input, 107x cheaper output). Sourced.
ReadEmbeddings Cost Calculator (2026): OpenAI, Voyage, Google, Cohere
Calculate the real cost of text embeddings in 2026 across OpenAI text-embedding-3, Voyage AI 3-large/3-lite, Google gemini-embedding, and Cohere embed-v4. Live per-1M-token prices, worked examples at 1M / 100M / 1B tokens, plus the hidden storage line nobody budgets for. Sourced.
ReadGPT-5 Cost Calculator (2026): GPT-5.5, GPT-5.5 Pro, GPT-5.4, 5.4-mini
Real $-per-call math for every GPT-5 model in June 2026. GPT-5.5 at $5/$30 per 1M, GPT-5.5 Pro at $30/$180, GPT-5.4 at $2.50/$15, GPT-5.4-mini at $0.50/$1.50. Batch (50% off), cache (90% off), worked examples at 1k, 100k, 1M, and a full agent loop. Sourced.
ReadGrok 4 API Cost Calculator (2026): Grok-4.3, 4.20, Grok-4 Fast
Calculate exactly what a Grok API call costs in 2026. Live per-1M-token prices for Grok-4.3, Grok-4.20, and Grok-4 Fast. The 90%-off cache-hit math, the $150/mo data-sharing free credit, and four worked $ examples at 1k, 100k, 1M calls and an agent loop. Sourced from xAI's docs.x.ai/docs/models.
ReadLlama 4 Cost Calculator (2026): Groq vs Together AI vs Replicate
Llama 4 is free to download. Inference is not. Per-1M-token prices for Llama 4 Scout and Maverick on Groq, Together AI, and Replicate — June 2026. Worked $ examples at 1k, 100k, 1M calls and a 5-turn agent loop. Self-host breakeven math included.
ReadMidjourney Cost Calculator (2026): Basic, Standard, Pro, Mega + Real $/Image
Calculate the real cost of Midjourney in 2026. Full plan table — Basic $10, Standard $30, Pro $60, Mega $120 — fast vs relax math, GPU-minute economics, $/image at scale, and the relax-vs-fast decision tree. Sourced + worked examples.
ReadMistral API Cost Calculator (2026): Large 2, Large 3, Medium 3.5, Small 4
Calculate exactly what a Mistral La Plateforme API call costs in 2026. Live per-1M-token prices for Mistral Large 2, Large 3, Medium 3, Medium 3.5, Small 4. Worked $ examples at 1k, 100k, 1M calls and a 5-turn agent loop. Mistral vs GPT-5.4 vs DeepSeek side-by-side. EU data-residency analysis. Sourced.
Reado1 / o3 Reasoning Cost Calculator (2026): The Thinking-Token Premium, Solved
Real $ math for OpenAI's o-series reasoning models in 2026. o3 ($2 / $8), o3-mini ($0.55 / $2.20), o1 ($15 / $60). Why a 200-token answer can cost 5-15x more than chat. Four worked examples, the 87% o1 to o3 price drop, and the decision tree for when reasoning models are actually worth it.
ReadOpenAI API Cost Calculator (2026): GPT-5.5, GPT-5.4, Batch + Cache
Calculate exactly what an OpenAI API call costs in 2026. Live per-1M-token prices for gpt-5.5, gpt-5.5-pro, gpt-5.4, gpt-5.4-mini, gpt-5.4-nano. Batch (50% off) and cached-input (90% off) discounts. Worked $ examples at 1k, 100k, and 1M calls. Formula box. Sourced.
ReadPerplexity Sonar API Cost Calculator (2026): Per-Token + Per-Request
Real $ math for the Perplexity Sonar API in 2026. Sonar, Sonar Pro, Sonar Reasoning Pro, Sonar Deep Research — token rates plus the per-request search fee that nobody else explains. Worked examples at 1k, 100k, 1M calls. Sourced from docs.perplexity.ai, June 2026.
ReadReplit Agent Cost Per Task (2026): Credits, Plans, Real $ Math
What does one Replit Agent task actually cost in 2026? Verified plan pricing (Starter $0, Core $25/mo, Pro $100/mo), credits-per-action billing explained, and 4 worked task examples from a $1.50 todo app to a $12 dashboard. Sourced from replit.com/pricing, June 2026.
ReadWindsurf (Devin Desktop) Cost Calculator 2026: Pro $20, Max $200, Teams $40
What Windsurf actually costs in 2026 after the Devin Desktop rebrand. Free, Pro $20, Pro Plus $35, Max $200, Teams $40/seat — verified June 2026. The credits-to-quota migration explained, 4 worked $ examples, Pro vs Pro Plus vs Max decision tree, and how it stacks vs Cursor and Copilot.
Read
Stop guessing your AI bill.
Digital Dashboard Hub turns your real spend across OpenAI, Anthropic, and Google into one live dashboard — usage, cost, budget alerts, model mix. 14 days free.
Try DDH free