Cost simulator
Estimate your monthly AI costs PrintChoose your models, describe how you will use them, and get a monthly estimate from each provider's published API rates. Turn on the subscription comparison to weigh it against a plan. If a rate is not published, the estimate reads Unavailable; it is never counted as free.
1
Choose your models Add up to four models and set how much of the workload each one handles. The shares always add up to 100%, so changing one adjusts the others. A single model handles the whole workload.
Reset to equal split Add a model Choose another model… Google · Gemini 4 Neon OpenAI · GPT-6.2 Astra Google · Gemini 4 Carbon OpenAI · GPT-6 Astra Anthropic · Claude Opus 5.5 OpenAI · GPT-6.2 Sol OpenAI · GPT-6.1 Sol OpenAI · GPT-6.1 Terra Google · Gemini 4 Argon Google · Gemini 4 Hydrogen Anthropic · Claude Fable 5.1 Anthropic · Claude Sonnet 5.5 Anthropic · Claude Opus 5 Anthropic · Claude Fable 5 Alibaba · Qwen4 Max OpenAI · GPT-6 Sol OpenAI · GPT-5.6 Sol Meta · Muse Wonder OpenAI · GPT-5.5 Pro Moonshot AI · Kimi K3.1 Pro DeepSeek · DeepSeek V4.1 Flash 1017 OpenAI · GPT-5.5 Tencent · Hy4 OpenAI · GPT-5.4 Pro Tencent · Hy4 preview · self-hosted Google · Gemini 3.8 Flash MiniMax · MiniMax M3.1 Flash Google · Gemini 3.7 Flash Anthropic · Claude Opus 4.8 OpenAI · GPT-5.6 Terra Thinking Machines Lab · Inkling 2 Small Anthropic · Claude Opus 4.7 (Adaptive) Moonshot AI · Kimi K3 xAI · Grok 4.6 Anthropic · Claude Haiku 5.5 OpenAI · GPT-5.2 Pro Anthropic · Claude Opus 4.7 OpenAI · GPT-5.4 Xiaomi · MiMo-V2.6-Pro Xiaomi · MiMo-V2.6-Flash StepFun · Step 5 Preview DeepSeek · DeepSeek V4 Pro Google · Gemini 3.6 Flash OpenAI · GPT-6 Luna Anthropic · Claude Sonnet 5 Google · Gemini 3.1 Pro xAI · Grok 4.5 DeepSeek · DeepSeek V4 Pro 0813 Google · Gemini 3 Pro OpenAI · GPT-5.6 Luna OpenAI · GPT-5.2 Meta · Muse Ember Anthropic · Claude Opus 4.6 Google · Gemini 3.5 Flash Meta · Muse Spark 1.2 Z.AI · GLM-5.2 OpenAI · GPT-5.3 Codex Mistral · Mistral Large 4 Moonshot AI · Kimi K2.5 (Reasoning) Thinking Machines Lab · Inkling 2 Mini Z.AI · GLM-5 (Reasoning) OpenAI · o3 Alibaba · Qwen3.5 397B (Reasoning) Meta · Muse Spark Open Anthropic · Claude Opus 4.5 OpenAI · GPT-5.2 Instant Moonshot AI · Kimi K2.6 OpenAI · GPT-5.3 Instant Anthropic · Claude Sonnet 4.6 OpenAI · GPT-5 OpenAI · GPT-5 (high) DeepSeek · DeepSeek V3.2 (Thinking) Google · Gemini 3 Flash OpenAI · GPT-5.1 MiniMax · MiniMax M3 Moonshot AI · Kimi K2.7 Code Z.AI · GLM-5.1 OpenAI · o3-pro OpenAI · GPT-5.4 mini xAI · Grok 4.20 Tencent · Hy3 · self-hosted Multiverse Computing · Quasar 438B OpenAI · GPT-5.2-Codex Tencent · Hy3 Preview · self-hosted Arcee AI · Trinity-Large-Preview Thinking Machines Lab · Inkling-Small Thinking Machines Lab · Inkling Google · Gemini 3.5 Flash-Lite Anthropic · Claude Sonnet 4.5 Alibaba · Qwen3.5 397B OpenAI · GPT-5.1-Codex-Max Prism ML · Ternary Bonsai 2 27B · self-hosted Z.AI · GLM-5 xAI · Grok 4.3 Meta · Muse Glimmer 30B · self-hosted Google · Gemini 2.5 Pro Z.AI · GLM-5-Turbo StepFun · Step 3.7 Flash StepFun · Step 3.5 Flash Moonshot AI · Kimi K2.5 Google · Gemini 2.5 Flash Anthropic · Claude 4.1 Opus OpenAI · GPT-5.4 nano OpenAI · o1 OpenAI · GPT-5.1-Codex Alibaba · Qwen3.5 Plus OpenAI · o4-mini (high) OpenAI · o3-mini DeepSeek · DeepSeek-R1 MiniMax · MiniMax M2.7 Mistral · Ministral 3 14B (Reasoning) Google · Gemini 3.1 Flash-Lite Mistral · Mistral Medium 3.5 128B Poolside · Laguna S 2.1 Anthropic · Claude 4 Sonnet Aion Labs · Aion-2.0 OpenAI · GPT-5 mini Upstage · Solar Pro 3 OpenAI · o1-preview Alibaba · Qwen3.7 Flash xAI · Grok 4.1 Fast xAI · Grok 3 Mini OpenAI · GPT-5 nano DeepSeek · DeepSeek V3.2 LiquidAI · LFM2.5-24B-A2B Microsoft · Phi-4 · self-hosted OpenAI · o1-pro xAI · Grok Code Fast 1 Z.AI · GLM-5V-Turbo MiniMax · MiniMax M2.5 Anthropic · Claude 3.5 Sonnet Z.AI · GLM-4.5 OpenAI · GPT-4.1 Z.AI · GLM-4.5-Air Mistral · Mistral Large 3 Alibaba · Qwen3.5 Flash OpenAI · GPT-OSS 120B · self-hosted DeepSeek · DeepSeek V3 Inception · Mercury 2.5 Arcee AI · Trinity-Large-Thinking Inception · Mercury 2 Mistral · Ministral 3 8B (Reasoning) Celeris · Celeris-1 Google · Gemini 1.5 Pro Anthropic · Claude 3 Haiku Cohere · Command A+ Mistral · Mistral Medium 3 Alibaba · Qwen3 Max Mistral · Ministral 3 3B (Reasoning) Moonshot AI · Kimi K2 Anthropic · Claude Haiku 4.5 Mistral · Mistral Small 4 OpenAI · GPT-4 Turbo Meta · Llama 4 Scout · self-hosted OpenAI · GPT-4.1 mini Anthropic · Claude 3 Opus OpenAI · GPT-4o OpenAI · GPT-OSS 20B · self-hosted OpenAI · GPT-4o mini Meta · Llama 4 Maverick · self-hosted Mistral · Ministral 3 14B NVIDIA · Nemotron 3 Super 120B A12B · self-hosted OpenAI · GPT-4.1 nano Mistral · Ministral 3 8B Mistral · Ministral 3 3B Celeris · Celeris-1 Decision Celeris · Celeris-1.1 Decision Anthropic · Claude Mythos 5 Anthropic · Claude Mythos 5.2 Anthropic · Claude Mythos Preview Cloudflare · Clef Cloudflare · Clef-Flash Cloudflare · Clef-Omni Cursor · Composer 2 Cursor · Composer 2.5 Liquid AI · d1-ultra Liquid AI · d1.2 Liquid AI · d1.2-ultra DeepSeek · DeepSeek V4 Flash 0731 DeepSeek · DeepSeek-R1-0528 Nace.AI · Drex Max Nace.AI · Drex Max v1.1 Nace.AI · Drex v1.5 Nace.AI · Drex v2.0 Cohere · Embed 5 Fast Cohere · Embed 5 Pro Fireworks · Ember-1 Sakana AI · Fugu Cyber Google · Gemini 2.5 Flash-Lite Google · Gemini 3.8 Live Google · Gemini 3.8 Live Extended Thinking Google · Gemini Decision 1 Google · Gemini Omni 1.1 Flash Google · Gemini Omni 1.2 Flash Fastino · GLiDE Fastino · GLiDE v1.2 Elorian · Glimpse Z.AI · GLM 5.3 FlashX Legal OpenAI · GPT Audio 1.5 OpenAI · GPT Realtime OpenAI · GPT Realtime 1.5 OpenAI · GPT Realtime 2 OpenAI · GPT Realtime 2.1 OpenAI · GPT Realtime 2.1 Mini OpenAI · GPT Realtime mini OpenAI · GPT-4.5 OpenAI · GPT-4o Audio OpenAI · GPT-4o mini Audio OpenAI · GPT-4o mini TTS OpenAI · GPT-6 Cyber OpenAI · GPT-6 Luna (Decisions) OpenAI · GPT-6.1 Terra Pro OpenAI · GPT-6.2 Astra Pro OpenAI · GPT-6.2 Sol Pro xAI · Grok 4.7 xAI · Grok Build 0.1 H Company · Holo3-122B-A10B H Company · Holo3.1-35B-A3B Tencent · Hy-MT2-30B-A3B · self-hosted Tencent · Hy4 flash Interfaze · Interfaze Beta TypeSafe AI · Jev 1.14 Moonshot AI · Kimi K3.1 Pro Highspeed Poolside · Laguna XS 2.1 Meituan · LongCat-2.5 Microsoft · MAI Code 2 Microsoft · MAI Code 2 Flash Microsoft · MAI-Code-1.1-Flash Inception · Mercury 3 Inception · Mercury Decide Inception · Mercury Edit 2 Inception · Mercury Voice Microsoft · Microsoft-Decision-1 Mistral · Mistral Small 4 (Reasoning) Meta · Muse Spark 1.3 Google · Nano Banana 2.1 OpenAI · o1-mini OpenAI · o4-mini Unbiased · Pareto 26.10 Preview Unbiased · Pareto 26.9 Particle AI · Particle-1 Perplexity · Perplexity Decider Flash v1.2 4B Perplexity · Perplexity Decider Flash v1.3 4B Perplexity · Perplexity Decider v1 27B Perplexity · Perplexity Decider v1.1 (27B) Perplexity · Perplexity Decider v1.2 27B Perplexity · Perplexity Decider v1.3 27B Pokee AI · Pokee-Isaac 28B Alibaba · Qwen3 32B Alibaba · Qwen3-VL-235B-A22B Instruct Cohere · Rerank 5 Fast Cohere · Rerank 5 Pro Ricursive Intelligence · Ripple-65B Preview Sakana AI · Sakana Fugu-Ultra v1.1 Sakana AI · Sakana Namazu ScaleDown · ScaleDown Classify ScaleDown · ScaleDown Compress ScaleDown · ScaleDown Extract ScaleDown · ScaleDown Summarize ByteDance · Seed 2.1 Turbo Upstage · Solar Pro 4 Subquadratic · SubQ Snap 1.1 Mixedbread · Toast 1 Paper Instruments · Ultramarine 1 Flash Add model2
Describe your monthly use Answer in your own terms and we convert it to tokens. These are your assumptions, not measured data, and each one is listed so you can check or change it.
Chat & writing People ask, draft and research in a chat window. Coding agent An agent reads and edits your repo, step by step (Codex, Claude Code, Cursor-style). Automation A script or workflow calls the model once per item: tickets, emails, documents.
Start from: Light Typical Heavy
≈ 220 chats/month × 8 model calls each × (2.6k in + 400 out) = 5.2M tokens/month
~800-token system prompt and ~80 tokens per thing you type Each reply resends the whole chat so far, so long threads cost more per message Answer length includes any hidden thinking tokens Prompt caching off: all input billed at the full rate How we estimate workloads Edit token counts directly (advanced) Your monthly estimate What the use in step 2 would cost each month at published API rates, split across the models in step 1. It is an estimate, not a bill.
Estimated API cost $64.20 / month · 5.21M modelled tokens
Published rates × your use and model shares.
Published rates Exactly what each source published. Nothing in this table is calculated.
Source prices for the selected models Item Component Source value Evidence Claude Fable 5.2 Input / 1M $8.00 Output / 1M $40.00 Cache read / 1M $0.20 Cache write / 1M $10.00
How the estimate adds up Calculated from your use and the published rates above.
Derived monthly cost per route Route Share Modelled tokens Monthly cost Status Claude Fable 5.2 100% 5.21M $64.20 Priced from source rates API total 100% 5.21M $64.20 Sum of the routes above
Copy scenario link Export CSV Print scenarioOne practical AI decision each week. AI into Production — by MonoMind AI Lab
Practical AI decisions, grounded in ModelChoice evidence. What to build, what to test, and what to leave human.