API costs
Downloads and updates run 12 months; the software you downloaded never expires. AI provider usage is separate.
Aon Pro Studio does not bundle hidden AI credits. You bring your own OpenAI API key for core AI features, and use free on-device voices at no charge or choose from four TTS voice providers - OpenAI, MiniMax, ElevenLabs, and Typecast - with MiniMax adding image-to-video and the high-quality production mode. Those providers bill your account directly based on the models and usage you choose.
Pricing references last reviewed: September 7, 2026.
What the $99 Pro license includes
- Aon Pro Studio desktop software for Windows 10/11 and Apple Silicon Macs (M1 or later, macOS 14 or later).
- Current Pro workflows, including the editor, One-Click, Full Automation, High-Quality production (narration stories, music videos, drama and ads), highlights, and batch render.
- Software downloads (re-downloads included) - guaranteed for 12 months from the purchase date.
- Product updates and future major version upgrades - guaranteed for 12 months from the purchase date.
- Software you download and activate - no time limit on use.
- Local software access through license-key activation.
What is not included
AI API usage fees are not included. OpenAI is required for core AI workflows. For higher-quality TTS you can add MiniMax, ElevenLabs, or Typecast, and MiniMax also adds image-to-video and the high-quality production mode. Provider accounts, keys, balances, model access, rate limits, and billing are managed by each provider.
Gemini is not currently supported.
Need to create the keys first? Follow the API key setup guide for step-by-step OpenAI, MiniMax, ElevenLabs, and Typecast instructions with official links.
What each model the app uses actually costs
These are the same tables as the Help → API Costs tab inside the app. They cover only the combinations the app actually sends, so there is nothing else to work out: images at medium quality, H3 as 5-15 second clips at 768P or 2K, H3 Max at 480P or 768P, and Hailuo 2.3 Fast billed per clip. Last checked against the providers' official price pages on October 1, 2026.
Image generation
The app generates every image at medium quality, portrait 1024×1536 or landscape 1536×1024. Both sizes cost the same.
| Model | Price | Quality | Value | Notes |
|---|---|---|---|---|
| OpenAI gpt-image-2.5-flareDefaultper image (1024×1536, medium) | token-based | Quality 5/5 | Value 3/5 | Released 2026-09-08. Better quality than gpt-image-2 and about twice as fast. Token rates match gpt-image-2 ($30 per 1M image output tokens), but per-image token use differs, so OpenAI publishes no fixed per-image price; gpt-image-2's $0.041 is the closest yardstick. |
| OpenAI gpt-image-2.5-sunburstBest qualityper image (1024×1536, medium) | token-based | Quality 5/5 | Value 2/5 | The precision-editing model from the same release. Best at following reference images, so it suits work that repeats a character; slower to generate. Billed like flare. |
| OpenAI gpt-image-2per image (1024×1536, medium) | $0.041 | Quality 4/5 | Value 2/5 | Accurate on-screen text, recommended for Motion Graphics. Org verification required. |
| OpenAI gpt-image-1.5per image (1024×1536, medium) | $0.050 | Quality 4/5 | Value 2/5 | Stable release; high quality without verification. |
| OpenAI gpt-image-1-miniper image (1024×1536, medium) | $0.015 | Quality 3/5 | Value 4/5 | Economical; weaker on detail and text but fine for batches. |
| MiniMax image-01Lowest priceper image | $0.0035 | Quality 3/5 | Value 5/5 | Cheapest, about 1/12 of gpt-image-2. Requires a MiniMax key. |
Video generation (image-to-video)
Hailuo 2.3 Fast is billed per clip; H3 and H3 Max per second. Both make 5-15 s clips at the same 768P rate. H3 renders 768P or 2K; H3 Max renders 480P or 768P. New Premium projects start on the cheapest H3 Max 480P; pick the model and quality in Settings or in each video step.
| Model | Price | Quality | Value | Notes |
|---|---|---|---|---|
| MiniMax Hailuo 2.3 FastDefaultLowest price768P · 6 s clip | $0.19 | Quality 3/5 | Value 5/5 | Fast, cheap default model. No audio. |
| MiniMax Hailuo 2.3 Fast768P · 10 s clip | $0.32 | Quality 3/5 | Value 4/5 | |
| MiniMax Hailuo 2.3 Fast1080P · 6 s clip | $0.33 | Quality 3/5 | Value 4/5 | 1080P is available for 6 s only. |
| MiniMax H3Best quality768P · per second | $0.08 | Quality 5/5 | Value 2/5 | Far better quality and generates audio. 5 s $0.40 to 15 s $1.20. One of the two models the Premium modes can use. |
| MiniMax H32K · per second | $0.13 | Quality 5/5 | Value 1/5 | About 1.6x the 768P rate. 5 s $0.65 to 15 s $1.95. Charged only when you pick 2K in Settings or the video step's quality box. |
| MiniMax H3 Max768P · per second | $0.08 | Quality 5/5 | Value 3/5 | A fast-generation version of H3 post-trained by fal.ai. Same 768P rate as H3, much faster. 5 s $0.40 to 15 s $1.20. No 2K. |
| MiniMax H3 Max480P · per second | $0.05 | Quality 4/5 | Value 4/5 | The cheapest H3-family option. 5 s $0.25 to 15 s $0.75. The default for new Premium projects. |
Voice generation (TTS)
1,000 Korean characters is roughly 3-4 minutes of speech; the examples below use that basis.
| Model | Price | Quality | Value | Notes |
|---|---|---|---|---|
| Free voices (Supertonic 3, on this computer)Lowest priceper minute of audio | Free | Quality 3/5 | Value 5/5 | Made on this computer with no API key and no charge. The voice model (about 400 MB) is downloaded once. Korean, English and Japanese, 10 voices. Works on Windows PCs and Apple silicon Macs. |
| OpenAI GPT voiceDefaultper minute of audio | about $0.015 | Quality 3/5 | Value 5/5 | Works with the OpenAI key alone. 1,000 chars ≈ $0.05. |
| MiniMax speech-2.8-turboper 1M characters | $60 | Quality 4/5 | Value 4/5 | 1,000 chars = $0.06. Natural Korean pronunciation and emotion. |
| ElevenLabs eleven_multilingual_v2Best qualitymonthly plan + credits | Varies by plan | Quality 5/5 | Value 2/5 | Most expressive premium voices. Pay-as-you-go beyond the plan's character allowance. |
| Typecast1 char = 1 credit | Free 15,000/mo · Lite $15/mo for 200,000 | Quality 4/5 | Value 3/5 | Strong Korean voice-actor voices. The free API plan gives 15,000 credits a month with 2 concurrent requests and no payment details. Lite bills $0.09 per 1,000 credits past the limit, charged in 100,000-credit increments. |
Scripts and analysis (text models)
The script model is chosen in Settings and covers scripts, titles, topic suggestions, highlight extraction, image prompts, and high-quality planning, video prompts and character analysis. Prices are input / output per 1M tokens; the per-script estimate assumes one short-form script (about 2,000 input and 1,000 output tokens). The reasoning models (all but gpt-5.2 and gpt-5.1) also bill reasoning tokens as output, so real usage runs a little higher. Every option is small next to images and video.
| Model | Price | Notes |
|---|---|---|
| OpenAI gpt-6-astraper 1M tokens (input / output) | $10.00 / $50.00 | Top reasoning model. The tightest scripts, and the highest cost per script. One short-form script ≈ $0.07. |
| OpenAI gpt-6.1-solper 1M tokens (input / output) | $2.00 / $10.00 | Released 2026-09-30. The successor to gpt-6-sol at the same price. A reasoning model that balances performance and value. One short-form script ≈ $0.01. |
| OpenAI gpt-6-lunaDefaultLowest priceper 1M tokens (input / output) | $0.10 / $0.50 | Released 2026-09-22. A light reasoning model that balances performance and value. The cheapest, and still writes good scripts. One short-form script ≈ $0.001. |
| OpenAI gpt-5.6-solper 1M tokens (input / output) | $4.00 / $20.00 | High-end reasoning model. Promotional price through 2026-11-21. One short-form script ≈ $0.03. |
| OpenAI gpt-5.6-terraper 1M tokens (input / output) | $2.00 / $12.00 | Balanced choice with good prose for the money. One short-form script ≈ $0.02. |
| OpenAI gpt-5.2per 1M tokens (input / output) | $1.75 / $14.00 | Stable successor to gpt-5.1. Writes fast without reasoning. One short-form script ≈ $0.02. |
| OpenAI gpt-5.1per 1M tokens (input / output) | $1.25 / $10.00 | The previous default. Writes fast without reasoning. One short-form script ≈ $0.01. |
| OpenAI GPT (AI Video Editor analysis)per 1-hour video (gpt-6.1-sol) | ≈ $0.20 | The AI Video Editor also uses the script model (estimate). Finding 10 Shorts adds about $0.10 and finding what to tighten for a concept a few cents. On gpt-6-luna all of it is around two cents; on gpt-6-astra the analysis is about $1. Prices are shown on the buttons first, and the amount actually spent is kept with the project. |
| OpenAI GPT (motion prompts)per 1M tokens (input / output) | $0.10 / $0.50 | Image-to-video motion prompts in Click and Motion modes and prompt summaries (gpt-6-luna). Under a cent per scene. |
| OpenAI transcriptionper minute of audio | $0.006 | Transcription for the AI Video Editor and subtitle sync. 1 hour = $0.36. |
What one job costs
Rough totals worked out from the rates above, using scene counts you can actually pick in the app.
| Task | Setup | Estimated cost |
|---|---|---|
| Click-mode short, 4 scenes (still images) | GPT script + 4 × gpt-image-2 + OpenAI voice | about $0.20 |
| Same short with image-01 | GPT script + 4 × image-01 + OpenAI voice | about $0.05 |
| Short, 4 scenes all as Hailuo 2.3 Fast 6 s video | image cost + 4 × 768P clips | image cost + $0.76 |
| Premium production, 4 scenes, 6 s H3 Max each (default) | 480P · 24 s × $0.05 | image cost + $1.20 |
| Same video with H3 at 768P | 768P · 24 s × $0.08 | image cost + $1.92 |
| AI Video Editor on a 1-hour video + 10 Shorts | transcription 60 min + GPT analysis and Shorts | about $0.65-0.75 |
ElevenLabs and Typecast combine monthly plans with pay-as-you-go credits, so what you actually pay depends on the plan. Check the official price pages below before relying on any figure here.
How to control API spend
- Start with a short test render before running a full batch.
- Use still images first, then apply MiniMax image-to-video only to selected scenes. The per-second H3 model shows a cost notice wherever you pick it, and high-quality production mode lets you generate only the scenes you select.
- In Music Video mode you can draw a still instead of generating video for a clip. Video is billed per second while a still costs a few cents, so it cuts the bill sharply — that clip shows a moving picture rather than real motion.
- Review scripts and prompts before generating images or voice.
- Monitor provider dashboards directly for usage, balance, model access, and rate limits.
- Keep API keys private and rotate them if you accidentally expose one.
Official pricing sources
Provider pricing changes over time. Verify the latest rates on the official provider pages before running large production batches.