Supported LLMS

Which large language models (LLMs) does Kelvin support?

The Kelvin Legal Data OS is designed to be LLM-agnostic. Where Kelvin relies on or integrates into large language models, it does so via a common interface that allows for the use of both local and remote LLMs. Kelvin's support is summarized in the table below.

Please note that the models below may be subject to additional terms of use or licensing restrictions. If you need assistance navigating these terms and their technical or strategic implications, please contact your named support representative or email us.

FamilyProviderModel(s)UsageLocation(s)
GPT
  • OpenAI
  • Microsoft (Azure)
  • text-davinci-003
  • gpt-3.5-turbo
  • gpt-3.5-turbo-16k
  • gpt-4
  • gpt-4-32k
  • gpt-4-1106-preview
Cloud
  • US (3)
  • EU (4)
  • AU (1)
PaLM2
  • Google
  • text-gecko
  • text-otter
  • text-bison
  • chat-bison
  • text-bison-32k
  • chat-bison-32k
  • text-unicorn
Cloud
  • US
  • Canada
  • Brazil
  • UK
  • Switzerland
  • EU (5)
  • India
  • Japan
  • HK
  • Taiwan
  • South Korea
  • Australia
  • Israel
ClaudeAnthropic
  • claude-instant-1
  • claude-2
Cloud
  • US
GPT-J/Neo(X)
  • Eleuther AI
  • gpt-j-6B
  • gpt-neo-125M
  • gpt-neo-1.3B
  • gpt-neo-2.7B
  • gpt-neox-20B
LocalN/A
T5
  • Google
  • t5-small
  • t5-base
  • t5-large
  • t5-3b
  • t5-11b
  • flan-t5-small
  • flan-t5-base
  • flan-t5-large
  • flan-t5-xl
  • flan-t5-xxl
LocalN/A
Dolly2
  • Databricks (Eleuther AI)
  • dolly-v2-7b
  • dolly-v2-12b
LocalN/A
MPT
  • Databricks*
  • mpt-7b
  • mpt-7b-instruct
  • mpt-7b-storywriter
  • mpt-30b
  • mpt-30b-instruct
LocalN/A
GPT4All
  • Nomic AI
  • gpt4all-j-v1.0
  • gpt4all-j-v1.1
  • gpt4all-j-v1.2
  • gpt4all-j-v1.3
LocalN/A
Llama2
  • Meta (Facebook)
  • llama-2-7b
  • llama-2-13b
  • llama-2-70b
  • Half-precision (-hf) and other quantized versions
LocalN/A

* Databricks acquired MosaicML acquired on June 26, 2023.