Choose a model for the work you need to doMe
Official APIInternationalVersion 1
Read image content, text, and tables and return text. File generation and numerical validation are separate steps.
Vision & OCR
API BASE URL
https://api.minimax.io/v1Actual API model ID
MiniMax-M3Request method / path
POST /chat/completionsAuthentication
Authorization: Bearer $API_KEYEdit request parameters
Edit JSON within the verified parameter ranges. Copy and export stop if types, required fields, or enum values are invalid.
# 先在当前终端设置环境变量：API_KEY, IMAGE_URL
# 下面只创建请求文件，不执行文本中的任何命令。
cat > request.json <<'MODEL_ATLAS_REQUEST'
{
"model": "MiniMax-M3",
"messages": [
{
"role": "user",
"content": [
{
"text": "请提取图片中的表格，输出结构化 JSON。",
"type": "text"
},
{
"type": "image_url",
"image_url": {
"url": "${IMAGE_URL}"
}
}
]
}
]
}
MODEL_ATLAS_REQUEST
# 替换 request.json 内的用户占位符后再运行请求。
curl --fail-with-body --request POST "https://api.minimax.io/v1/chat/completions" \
--header "Content-Type: application/json" \
--header "Authorization: Bearer ${API_KEY}" \
--data-binary @request.json