OpenAI
OpenAI itself is supported, but any openai api compatible services are also supported by the same provider in this plugin. See Ollama as an example.
Example config:
ai:
# Configure one or more text models
textModels:
- name: ollama-phi-2
modelName: phi-2
provider: openai
baseUrl: http://localhost:11434/v1
requireAuth: false
- name: gpt-4o
provider: openai
modelName: gpt-4o
- name: gpt-4o-mini
provider: openai
modelName: gpt-4o-mini
- name: gpt-4-turbo
provider: openai
modelName: gpt-4-0125-preview
- name: gpt-4-vision-preview
provider: openai
modelName: gpt-4-vision-preview
- name: gpt-3-turbo
provider: openai
modelName: gpt-3.5-turbo-0125
- name: Name to use inside of silverbullet for this model
- provider: openai
- modelName: One of the models listed on OpenAI’s list of models if using OpenAI. If not using OpenAI, follow the API provider’s documentation.
- baseUrl: Only needed if not using the official OpenAI models and api.
- requireAuth: If false, the Authorization headers will not be sent. Needed as a workaround for some CORS issues with Ollama.
- secretName: Name of secret to look for in SECRETS. If not set,
OPENAI_API_KEY
is used.
Cost (OpenAI)
While this plugin is free to use, OpenAI does charge for their API usage. Please see their pricing page for cost of the various apis.
Due to price and performance, gpt-4o is currently the recommended model if using OpenAI.
As of 2024-07, here's a rough idea of what to expect:
- Dall-E image generation, HD 1024x1024; $0.080 per image
- GPT-4o; $0.0050 per 1K input tokens, $0.0150 per 1K output tokens
- GPT-4o-mini; $0.000150 / 1K input tokens, $0.000600 / 1K output tokens
- GPT-4-turbo; $0.01 per 1k input tokens, $0.03 per 1k output tokens
- GPT-3.5-turbo; $0.0005 per 1k input tokens, $0.0015 per 1k output tokens
- Per the above pricing page, a rough estimate is that 1000 tokens is about 750 words
Recommended models
In general, gpt-4o is recommended if using OpenAI. gpt-4o-mini is a good alternative if you are looking for a more cost effective model that still provides good results. There's almost no reason to use gpt-3.5-turbo, gpt-4-turbo, or gpt-4 anymore.