API Documentation
01 / Get started
Get started
Complete these steps in order. You can test the connection as soon as you have an API key and a model name.
- 01
Register or sign in
Create an account or sign in to an existing account before using the API.
- 02
Balance
Keep enough balance before production traffic
- 03
Create an API key
Open the console, create a key for your application, and keep the full value private.
- 04
Choose a model
Compare available models, then select an exact model ID that your account can use.
- 05
Send your first request
Replace the key and model placeholders in one of the examples below and run it.
- 06
Usage and billing
Review usage logs, quota, costs, and payment status from the console.
Console workflow
Review usage logs, quota, costs, and payment status from the console.
Next step
After creating a key, continue with the API basics and copy the example that matches your client.
User guide
Loading...
User guide
Loading...
User guide
Loading...
User guide
Loading...
User guide
Loading...
User guide
Loading...
User guide
Loading...
02 / API
API Basics
Use the service address, API key, and model ID below as the three values shared by every client.
Base URL
https://api.flowvision.techUse the service address shown here. Do not include the documentation page path.
Authentication
Authorization: Bearer sk-your-api-keyCreate the key in the console and send it with every API request.
Model name
your-modelUse the exact model ID returned by the model list.
API Basics
Available Models
Use this request to see the model IDs available to your API key.
curl https://api.flowvision.tech/v1/models \
-H "Authorization: Bearer sk-your-api-key"API Basics
API Requests
Use the format that matches your client. Enable only the upstream models and protocols configured by your administrator.
OpenAI Chat Completions
/v1/chat/completionsOpenAI Responses
/v1/responsesAnthropic
/v1/messagesGemini
/v1beta/models/{model}:generateContentImage
/v1/images/generationsVideo
/v1/videosRerank
/v1/rerankEmbeddings
/v1/embeddingsAudio
/v1/audio/speech03 / Requests
Request formats
Choose the format supported by your client. All formats use the same FlowVision key.
For Anthropic clients, x-api-key is accepted. Bearer authentication is also supported.
Request formats
OpenAI Chat Completions format
Use a messages array for standard chat requests.
curl https://api.flowvision.tech/v1/chat/completions \
-H "Authorization: Bearer sk-your-api-key" \
-H "Content-Type: application/json" \
-d '{
"model": "your-model",
"messages": [
{"role": "user", "content": "Hello"}
]
}'Request formats
OpenAI Responses format
Use input for Responses requests, including Codex.
curl https://api.flowvision.tech/v1/responses \
-H "Authorization: Bearer sk-your-api-key" \
-H "Content-Type: application/json" \
-d '{
"model": "your-model",
"input": "Hello",
"max_output_tokens": 128
}'Request formats
Anthropic Messages format
Use messages together with max_tokens for Claude clients.
curl https://api.flowvision.tech/v1/messages \
-H "x-api-key: sk-your-api-key" \
-H "anthropic-version: 2023-06-01" \
-H "Content-Type: application/json" \
-d '{
"model": "your-model",
"max_tokens": 128,
"messages": [
{"role": "user", "content": "Hello"}
]
}'For Anthropic clients, x-api-key is accepted. Bearer authentication is also supported.
04 / Models
Model usage by scenario
Use the guide below to choose the right API shape for conversational and video models.
Use the FlowVision API key created in the console. End users do not need upstream provider credentials.
Model usage by scenario
Conversational models
Use Chat Completions for ordinary conversational requests.
Recommended endpoint
POST /v1/chat/completionsExample model ID
deepseek-v4-flash-ga-260731curl https://api.flowvision.tech/v1/chat/completions \
-H "Authorization: Bearer sk-your-api-key" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4-flash-ga-260731",
"messages": [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Explain what FlowVision does in one sentence."}
],
"temperature": 0.7
}'The model ID is an example. Replace it with an id returned by /v1/models.
Use /v1/responses for Responses clients such as Codex CLI.
Model usage by scenario
Video models · Seedance
Video generation is asynchronous. Save the task ID and poll until the task is complete.
Recommended endpoint
POST /v1/video/generationsExample model ID
doubao-seedance-2-0-260128curl https://api.flowvision.tech/v1/video/generations \
-H "Authorization: Bearer sk-your-api-key" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seedance-2-0-260128",
"prompt": "A ginger cat walking through a sunny courtyard, cinematic camera movement",
"duration": 5,
"ratio": "16:9",
"resolution": "720p",
"output_format": "mp4",
"generate_audio": true,
"watermark": false
}'Request parameters
- prompt
- Prompt text that describes the video to generate.
- duration
- Duration in seconds; supported values are 4–15 for most Seedance models and 4–30 for Seedance 2.5.
- ratio
- Output ratio, such as 16:9 or 9:16.
- resolution
- Output resolution; 720p is a safe default.
- output_format
- Output format; mp4 is supported by all Seedance models.
- generate_audio
- Whether to generate audio; defaults to true.
- watermark
- Whether to add a watermark; defaults to false.
For image-to-video, add references with an HTTPS image_url; video_url and audio_url are supported when the selected model allows them.
curl https://api.flowvision.tech/v1/video/generations/task_xxx \
-H "Authorization: Bearer sk-your-api-key"
curl https://api.flowvision.tech/v1/videos/task_xxx/content \
-H "Authorization: Bearer sk-your-api-key" \
-o output.mp4Use the returned task ID to check progress. A completed task can be downloaded from the content endpoint.
05 / Client setup
Codex CLI
Use the OpenAI Responses protocol with FlowVision as a custom provider.
The API key used here is issued by FlowVision. It is not an upstream provider key.
1. Add the provider
Save this configuration in your Codex config file: ~/.codex/config.toml
model = "your-model"
model_provider = "flowvision"
[model_providers.flowvision]
name = "FlowVision-Api"
base_url = "https://api.flowvision.tech/v1"
env_key = "FLOWVISION_API_KEY"
wire_api = "responses"2. Set the API key
Set the environment variable before starting Codex.
export FLOWVISION_API_KEY="sk-your-api-key"
codexUse the exact model ID returned by the model list. The provider URL in the configuration includes the API version path.
Add models without replacing existing models
Merge available models into your Codex catalog, back up your configuration, then restart Codex. Model access depends on your API key and channel support.
model_catalog_json = "/absolute/path/flowvision-models.json"06 / Client setup
Claude Code
Claude Code uses the Anthropic Messages format and reads its connection settings from environment variables.
Base URL
https://api.flowvision.techDo not add the API version suffix to this value.
API key
ANTHROPIC_API_KEYUse the key created in Console → API Keys.
Model name
ANTHROPIC_MODELSet this to an exact model ID available to your key.
Set these variables before starting Claude Code.
export ANTHROPIC_BASE_URL="https://api.flowvision.tech"
export ANTHROPIC_API_KEY="sk-your-api-key"
export ANTHROPIC_MODEL="your-model"
claude07 / Help
Troubleshooting
If a request fails, check the status code, the model name, and the key permissions first.
401Authentication failed
The key is missing or invalid. Check the Authorization header or create a replacement key.
403Permission denied
Your key is valid, but your account or key cannot use this model or route.
404Endpoint not found
Check the service address and endpoint path, and make sure the address does not include the docs page.
429Rate limit or insufficient balance
You have reached a rate limit or your available balance is insufficient. Check the console before retrying.
5xxUpstream unavailable
The selected model or its service is unavailable. Check the model status and try again later.
08 / Security
Security
A small amount of key hygiene prevents most avoidable access problems.
Treat your API key like a password.
Do not place the key in browser code or public client applications.
If the key is exposed, revoke it in Console → API Keys and create a replacement.
Use separate keys for different applications so you can revoke one without interrupting the others.
Ready to test? Open the API key page, copy a model ID, and run the first example.
Open API Keys