OpenFilmPricing
OpenFilm is free. Making, editing and rendering films costs nothing, and your agent can use any service with its own key. You pay OpenFilm only for what you use through an OpenFilm account: the OpenFilm agent in the desktop app, and generated voice, music, images and video.
One rate: the real price × 1.1
Everything paid through OpenFilm costs what it cost us, times 1.1. The agent's models run through OpenRouter, and each answer says what it cost; media services are billed the same way, by what each call used. Nothing else is added: no minimum per call, and no extra charge per model.
Why 0.1: OpenFilm is open source under the MIT license, so that anyone can use it, change it and build on it; that comes first. The 0.1 pays the fees OpenRouter and the other services charge on what we buy and the servers that pass your requests on, and what is left keeps the people making OpenFilm working on it. The rate is the same for everyone and written here, so you can always check it.
Plans buy credits at 500 per US dollar, so $1 of real cost is 550 credits.
The cheapest way to use OpenFilm
- A subscription you already have. Codex (with a ChatGPT plan) and Claude Code (with a Claude plan) work in the desktop app, signed in with your own account. For many hours of work with a strong model, a subscription costs the least.
- Your own key. In the desktop app, Settings → Agents → Your own key runs an OpenAI or Anthropic API key, or any service with the same API. You pay the service directly; OpenFilm adds nothing.
- The OpenFilm agent. Nothing to set up: pick a model and go, paid in credits at the real price × 1.1. As OpenFilm grows and buys in larger amounts, the prices we pay fall, and yours fall with them.
The OpenFilm agent's models
Every call runs on the model maker's own service through OpenRouter, never on another host. Prices are per million tokens, as OpenRouter lists them; a call costs what OpenRouter charged for it × 1.1.
| Model | Input | Output | Cache read | Cache write |
|---|---|---|---|---|
| DeepSeek V4.1 Flash (default) | $0.30 | $1.20 | $0.006 | free |
| MiMo V2.6 Flash | $0.14 | $0.28 | $0.0028 | free |
| MiMo V2.6 Pro | $0.435 | $0.87 | $0.0036 | free |
| GPT-6.1 Sol | $2 | $10 | $0.10 | $2.50 |
| Claude Sonnet 5.5 | $2 | $10 | $0.20 | $2.50 |
| Claude Opus 5.5 | $4 | $20 | $0.20 | $5 |
| GPT-6 Astra | $10 | $50 | $1 | $12.50 |
| Claude Fable 5.1 | $10 | $50 | $0.25 | $12.50 |
Every model keeps what a conversation repeats (its instructions, the turns so far) in a cache: the next turn reads it back at the cache price instead of paying for it again. OpenAI and Anthropic charge once for writing it; the others do not.
- DeepSeek has busy hours: weekdays 01:00–04:00 and 06:00–10:00 UTC, at the prices above. The rest of the time, and all weekend, it costs half: $0.15 input, $0.60 output, $0.003 cache read.
- OpenAI's models double every price once a single request passes 272,000 tokens of input.
- A maker's price change applies from the next call.
In credits, $1 is 550: DeepSeek V4.1 Flash is 165 credits per million input tokens and 660 per million output tokens; Claude Fable 5.1 is 5,500 and 27,500.
Media
Voice, sound effects, music, images, video and transcription from openfilm get or Studio follow the same rule: what the service charged for the call × 1.1. Each kind of media has its models, named for what they are; choose one in Studio (Settings → Models) or for one run with openfilm get … --model <id>. The first of each kind is the default and the cheapest. A free account has the models marked free; any plan has them all.
| Makes | Model | Real price | Credits, at most | Free account |
|---|---|---|---|---|
| Voice-over | Inworld TTS-2 Flash | $10 per 1M characters | 1 per 100 characters | yes |
| Voice-over | Cartesia Sonic 3.6 | $50 per 1M characters; Pro voices 1.5× | 3 per 100 characters (5 for a Pro voice) | |
| Voice-over | ElevenLabs Eleven v4 | $80 per 1M characters; some library voices 2× or 3× | 5 per 100 characters (up to 14 for a 3× voice) | |
| Sound effects | OpenFilm library | free | free | yes |
| Sound effects | ElevenLabs Sound Effects | $0.12 per minute | 2 per second | |
| Music | OpenFilm library | free | free | yes |
| Music | ElevenLabs Music v2 | $0.15 per minute | 2 per second | |
| Images | GPT Image 2.5 Flare | by tokens: $5 per 1M text in, $8 per 1M image in, $30 per 1M out | 3 per image | yes |
| Images | GPT Image 2.5 Sunburst | the same tokens, at high quality | 91 per image | |
| Video | Seedance 2.0 Mini (720p, 4–15 s) | $0.0756 per second | 42 per second | yes |
| Video | MiniMax Hailuo 3 (2K, 5–15 s) | $0.13 per second, $0.04 per reference image past 5 | 72 per second | |
| Video | Seedance 2.5 (720p, 4–30 s) | $0.2311 per second | 128 per second | |
| Transcription | ElevenLabs Scribe v2 | $0.22 per hour | 3 per minute | yes |
| Subtitle translation | DeepSeek V4.1 Flash | $0.30 / $1.20 per 1M tokens in / out (half outside busy hours) | 2 per 1,000 characters | yes |
| Image search | Google Images (Serper) | $1 per 1,000 queries | 1 per query | yes |
| Web search | Parallel | $1 per 1,000 queries | 1 per query | yes |
A run is charged what it measured, so it usually costs less than the most shown here: the column rounds up to whole credits per unit, and a run's real cost is counted exactly. A voice that costs more than its model's base price says so where voices are listed, and is charged at its own price. A voice-over runs only on the model you chose: if it fails there, it fails and costs nothing, and is never made by another vendor at another price. Looking up voices and fonts, and reading web pages, are free.
Credits
Credits come with a plan, and every plan buys them at the same rate. Credits you do not use carry over for one month. A call that fails is not charged.