AI Agent Multi-Modal Cost Calculator
Estimate monthly spend for agents that read images, audio, and video alongside text. Compare OpenAI, Anthropic, and Google. Edit rates to match your negotiated pricing.
Workload
Rates are per 1M tokens or per unit. Last updated: 2026-07-07. See notes.
| Provider / Model | Text input | Cached | Text output | Image | Audio / min | Video / min | Monthly cost | Best for |
|---|---|---|---|---|---|---|---|---|
| OpenAI GPT-4o gpt-4o-2024-08-06 | — | Best balance of vision, audio, and text quality | ||||||
| OpenAI GPT-4o-mini gpt-4o-mini-2024-07-18 | — | Cheapest vision + text workloads | ||||||
| Anthropic Claude 3.5 Sonnet claude-3-5-sonnet-20241022 | — | High-quality vision reasoning with large context | ||||||
| Anthropic Claude 3 Haiku claude-3-haiku-20240307 | — | Fast, cheap vision tasks | ||||||
| Google Gemini 1.5 Pro gemini-1.5-pro-002 | — | Long-context video and audio analysis | ||||||
| Google Gemini 1.5 Flash gemini-1.5-flash-002 | — | Cheapest long-context multimodal agents |
Cheapest option
Enter a workload to see the cheapest provider.
Media cost dominates for image, audio, and video agents. Compare the per-request breakdown below.
Cost breakdown per 1,000 requests
Enter a workload to see the breakdown.
Frequently asked questions
Which provider is cheapest for image-heavy agents?
Google Gemini 1.5 Flash and OpenAI GPT-4o-mini are usually the cheapest for vision workloads. Claude 3 Haiku is also competitive for high-quality image reasoning. Use your actual image count and resolution to compare.
How is audio/video cost estimated?
Audio is priced per minute of input and video is priced per minute of input (or per frame, depending on provider). The calculator uses representative per-minute rates; edit them to match the latest provider pricing.
Can I change the provider rates?
Yes. Every rate is editable inline. Click a cell, type a new value, and the calculator updates instantly. Use this when providers change pricing or you have negotiated rates.
What is not included?
Output media generation (image/audio generation), storage, bandwidth, vector DB, retries, fallback, and human review are not included. Pair this with the Agent API Pricing Comparator, Voice Agent Cost Calculator, and Document Processing Cost Calculator for full TCO.
Related calculators
Multimodal pricing is approximate and changes frequently. Use the editable rates below to match current provider pages. Video pricing is often billed per minute or per frame; check the provider's latest vision/audio pricing before committing.