API access to X1 series of models
We are excited to announce that you can now access the X1 series of models through our API! Here is a brief introduction to each of the models and their intended use cases:
X1-large: Our best model - scores 8.1 on the MT Bench, surpassing Claude 2. This model excels at general conversational tasks and information extraction. It has a context length of 8192.
X1-large-32k: Finetuned Llama 2 70b with 32768 token context length. If you have large system prompts you want to test out - this model is the best.
X1-CodeChat: Finetuned CodeLlama-34b. We chat-tuned the CodeLlama-34B model so that users will be able to interact with it better. This model excels at any coding task and works perfectly as a Coding assistant
X1-7b: Finetuned Mistral-7B. This model has a context length of 8192 along with function calling feature. Scores 7.82 on MT Bench and is the perfect model if the tasks are not extremely reasoning based.
Here are the steps to access our models through OpenAI's library:
Sign up on our console.
Add a payment method under the billing section - we partner with Stripe for this, and it is fully secure.
Create an API Key under the API Keys section.
You can use the X1 series of models instead of OpenAI's models with only two changes in your current OpenAI setup by changing the API Key and adding a new line below the openai's API Key:
openai.api_key = "GIGA ML's API KEY HERE"
openai.api_base = "https://inference.gigaml.com/v1"
You can access one of the two models: "X1-large", "X1-7b"
Here is a sample code for your reference:
Here is the pricing scheme for various models through our API:
X1-large: $0.005 per 1k prompt tokens, $0.01 per 1k completion tokens
X1-7b: $0.001 per 1k prompt tokens, $0.002 per 1k completion tokens
Please keep in mind that the above API access is not meant to be used in high-load production setups. If you want us to host your production setup or want to deploy our models on-prem: please contact us at email@example.com