Image Input via API: New Feature Now Available on Straico
We know that many of you have been asking for image input via API, and now it is a reality. This new API feature enables you to automate tasks that require image analysis using leading vision LLMs such as Claude 3.7 Sonnet, Llama 4 Maverick, or Gemini 2.5 Pro.
How to Use It
When calling the /v1/prompt/completion endpoint, you can now include an optional parameter called "images". This parameter should contain an array of strings, each representing the URL of an image.
Be sure to select only image-input-compatible models. To verify compatibility, use the /v1/models information endpoint and check that the model’s "features" array contains the string "image_input".
Restrictions
To prevent costly errors and maintain sustainability, image input is disabled for models costing more than 10 coins per 100 words. This restriction applies to the following models:
- OpenAI: o1 High Reasoning
- Perplexity: Sonar Deep Research
- Anthropic: Claude 3 Opus
- OpenAI: o1
- OpenAI: o3
- Google: Gemini Pro 2.5 Preview
Additionally, models released after this announcement that cost over 10 coins per 100 words will also have image input disabled.
Pricing Examples
Below are pricing examples for some popular models under the new capped Pricing per Message system:
-
Claude 3.7 Sonnet: Cost ranges from 5 coins (for very short prompts) to 75 coins (for longer ones).
Example:- 400 words cost approximately 21 coins.
- 800 words cost approximately 54 coins.
- For inputs above 1000 words, the cost is capped at 75 coins.
-
GPT-4o: Cost ranges from 3.3 coins (for very short prompts) to 49.5 coins (for longer ones).
Example:- 400 words cost approximately 14 coins.
- 800 words cost approximately 35 coins.
- For inputs above 1000 words, the cost is capped at 49.5 coins.
Why the Change?
This update is designed to promote effective API usage and provide a fair pricing structure, balancing the benefits of advanced image analysis with cost control.
- Prevents Misuse: The previous fixed pricing sometimes led to excessive costs. This new model encourages prudent usage.
- Cost-Effective: Short prompts benefit from lower costs, while longer inputs have a predictable cap.
- Sustainable: By limiting high-cost usage, we help maintain the long-term sustainability of our platform.
Key Benefits
- Seamless integration of image input for enhanced task automation.
- Flexible pricing ensures affordability regardless of prompt length.
- Prevents unexpected costs, making budgeting more predictable.
- Supports advanced models and diverse API workflows.
For further details, please consult our full documentation.
