Free apps. Pay only for what you use.
No subscription. No per-seat fees. You buy credits, and the AI’s work is billed per token at the prices below. New accounts start with free credits.
One bill.
Two simple fees.
Every model is available through one account, with no provider subscriptions and no API keys. Prices start from each provider’s public list price.
Provider list price
The tables below show each provider’s public list price by default, the same numbers you would see going direct. Every fee starts from there.
In the apps
When Chat, Create, or Admin calls a model directly, a 30% platform fee is added to the list price (1.3× list).
Through a harness
When a model runs inside Connect, Build, or Forge, including workflows on the Forge runtime, a 20% harness fee is added (1.5× list). The fees never stack. The fee includes the dedicated secure environment your agents and workflows run in, with no separate hosting bill, and a harness engineered to minimize token usage.
Every model. One account.
Text generation
USD per 1M tokens| Model | Input | Output |
|---|---|---|
| OpenAI GPT-5.5 Pro over 200k tokens: ×2 input / ×1.5 output | $30.00 | $180.00 |
| OpenAI o3 Pro | $20.00 | $80.00 |
| OpenAI o3 Deep Research | $10.00 | $40.00 |
| OpenAI GPT-5.5 over 200k tokens: ×2 input / ×1.5 output · cached input: ×0.1 | $5.00 | $30.00 |
| Anthropic Claude Opus 4.8 cache read: ×0.1 · cache write 1 h: ×2 · cache write 5 min: ×1.25 | $5.00 | $25.00 |
| Anthropic Claude Sonnet 4.6 cache write 5 min: ×1.25 · cache read: ×0.1 · cache write 1 h: ×2 | $3.00 | $15.00 |
| OpenAI GPT-5.4 over 272k tokens: ×2 input / ×1.5 output · cached input: ×0.1 | $2.50 | $15.00 |
| Amazon Nova 1 Premier | $2.50 | $12.50 |
| Google Gemini 3.1 Pro over 200k tokens: ×2 input / ×1.5 output · cached input: ×0.1 | $2.00 | $12.00 |
| Cohere Command A | $2.50 | $10.00 |
| Cohere Command R+ | $2.50 | $10.00 |
| Google Gemini 3.5 Flash cached input: ×0.1 · audio input: ×2.333 | $1.50 | $9.00 |
| OpenAI o3 | $2.00 | $8.00 |
| OpenAI o4 Mini Deep Research | $2.00 | $8.00 |
| Qwen Qwen 3.7 Max | $2.50 | $7.50 |
| Mistral Mistral Medium 3.5 | $1.50 | $7.50 |
| Anthropic Claude Haiku 4.5 cache write 5 min: ×1.25 · cache write 1 h: ×2 · cache read: ×0.1 | $1.00 | $5.00 |
| OpenAI GPT-5.4 Mini cached input: ×0.1 | $0.75 | $4.50 |
| OpenAI o4 Mini | $1.10 | $4.40 |
| Qwen Qwen 3.6 Plus | $0.50 | $3.00 |
| xAI Grok 4.3 | $1.25 | $2.50 |
| Amazon Nova 2 Lite | $0.30 | $2.50 |
| Mistral Devstral 2 | $0.40 | $2.00 |
| Mistral Mistral Large 3 | $0.50 | $1.50 |
| Google Gemini 3.1 Flash Lite cached input: ×0.1 · audio input: ×2 | $0.25 | $1.50 |
| Qwen Qwen 3.6 Flash | $0.25 | $1.50 |
| OpenAI GPT-5.4 Nano cached input: ×0.1 | $0.20 | $1.25 |
| DeepSeek DeepSeek 4 Pro | $0.435 | $0.87 |
| Cohere Command R | $0.15 | $0.60 |
| Meta Llama 4 Maverick | $0.15 | $0.60 |
| Mistral Mistral Small 4 | $0.15 | $0.60 |
| Meta Llama 3.1 70B | $0.40 | $0.40 |
| DeepSeek DeepSeek 3.2 | $0.25 | $0.38 |
| Google Gemma 4 31B | $0.12 | $0.37 |
| Google Gemma 4 26B A4B | $0.06 | $0.33 |
| Meta Llama 3.3 70B | $0.10 | $0.32 |
| Meta Llama 4 Scout | $0.08 | $0.30 |
| Mistral Ministral 3 14B | $0.20 | $0.20 |
| DeepSeek DeepSeek 4 Flash | $0.10 | $0.20 |
| Mistral Ministral 3 8B | $0.15 | $0.15 |
| Cohere Command R7B | $0.038 | $0.15 |
| Mistral Ministral 3 3B | $0.10 | $0.10 |
| Meta Llama 3.1 8B | $0.02 | $0.03 |
Image generation
USD per 1M tokens| Model | Input | Output |
|---|---|---|
| Google Nano Banana Pro text output: ×0.1 | $2.00 | $120.00 |
| Google Nano Banana 2 text output: ×0.05 | $0.50 | $60.00 |
| OpenAI GPT Image 2 text input: ×0.625 | $8.00 | $30.00 |
| OpenAI GPT Image 1 Mini text input: ×0.8 | $2.50 | $8.00 |
Video generation
USD per second of video| Model | Input | Output |
|---|---|---|
| ByteDance Seedance 2 480p: ×0.5 output · 720p: ×0.5 output | — | $0.40 |
| Google Veo 3.1 | — | $0.40 |
| Alibaba HappyHorse 1 480p: ×0.9 output · 720p: ×0.9 output | — | $0.30 |
| Runway Gen4.5 | — | $0.12 |
| Google Veo 3.1 Fast 1080p: ×1.2 output | — | $0.10 |
| Runway Gen4 Turbo | — | $0.05 |
Text to speech
USD per 1,000 characters| Model | Input | Output |
|---|---|---|
| ElevenLabs Eleven Multilingual v2 | — | $0.10 |
| ElevenLabs Eleven v3 | — | $0.10 |
| ElevenLabs Eleven Flash v2.5 | — | $0.05 |
| ElevenLabs Eleven Turbo v2.5 | — | $0.05 |
| OpenAI TTS 1 HD | — | $0.03 |
| OpenAI GPT-4o Mini TTS | — | $0.02 |
| OpenAI TTS 1 | — | $0.015 |
Music generation
USD per minute of audio| Model | Input | Output |
|---|---|---|
| ElevenLabs Eleven Music | — | $0.15 |
Transcription
USD per 1M tokens| Model | Input | Output |
|---|---|---|
| OpenAI GPT-4o Transcribe | $2.50 | $10.00 |
| OpenAI Whisper 1 per minute of audio | — | $0.006 |
| OpenAI GPT-4o Mini Transcribe | $1.25 | $5.00 |
Realtime voice
USD per 1M tokens| Model | Input | Output |
|---|---|---|
| OpenAI GPT Realtime 1.5 text input: ×0.125 · text output: ×0.25 | $32.00 | $64.00 |
| OpenAI GPT Realtime 2 text input: ×0.125 · text output: ×0.375 | $32.00 | $64.00 |
| OpenAI GPT Mini Realtime text output: ×0.12 · text input: ×0.06 | $10.00 | $20.00 |
Web search
USD per 1M tokens| Model | Input | Output |
|---|---|---|
| Perplexity Sonar Pro | $3.00 | $15.00 |
| Perplexity Sonar Deep Research | $2.00 | $8.00 |
| Perplexity Sonar Reasoning Pro | $2.00 | $8.00 |
| Perplexity Sonar | $1.00 | $1.00 |
Embeddings
USD per 1M tokens| Model | Input | Output |
|---|---|---|
| OpenAI Text Embedding 3 Large | $0.13 | — |
| OpenAI Text Embedding 3 Small | $0.02 | — |
Pricing,
answered.
Are the apps really free?
Yes. Every Crewdle app, Admin, Chat, Connect, Create, Build, and Forge, is free to use, with no subscription and no per-seat fee. You buy credits and pay only when the AI does work, at the model prices on this page. New accounts start with free credits.
What is a token?
Tokens are how AI models measure work. A token is roughly three quarters of a word: about 750 words make 1,000 tokens. A typical email runs a few hundred tokens, so 1M tokens covers a lot of everyday work. Input tokens are what you send to the model; output tokens are what it writes back.
How is the price of a model call calculated?
Every price starts from the provider’s public list price, shown by default in the tables. A model call from the apps adds a 30% platform fee (1.3× list). When the model runs through one of our harnesses, Connect, Build, or Forge, including workflows on the Forge runtime, a 20% harness fee is added instead of stacking (1.5× list). One bill, no separate provider subscriptions, no API keys to manage.
What does the harness fee include?
Two things. First, the dedicated secure environment your agents and workflows run in: we host and operate it, so there is no server to rent and no separate hosting bill, and idle agents cost nothing. Second, the harness itself, which enforces best practices like context management and caching to minimize the tokens each job consumes.
Do prices change?
Model prices follow the providers’ list prices, so when a provider lowers a price or releases a better model, you benefit right away. Prices on this page are in USD and reflect the current catalog.
What about services pricing?
Our services, AI classes, configured agents, static websites, turn-key builds, and the bank of hours, have fixed prices listed on the services page. The day-to-day runtime of anything we build for you is billed as usage at the model prices on this page.
Start for free.
Today.
No subscription. No credit card. Pay only when the AI works. If you're a real business, we'll throw in a 30-minute demo with a human.
