With Carbon being acquired by Perplexity last week, we've been talking to many ex-Carbon customers about transitioning to Graphlit.

We've put together a comparison sheet of features and data connectors to compare between many of the leading RAG-as-a-Service providers, so you can make your own choice.

We're limiting this analysis to API-first RAG-as-a-Service vendors, where you can ingest your or your customers' data into a platform, and perform semantic search and RAG over top of the retrieved content. We aren't comparing any UI features across the vendors.

Data came from vendor websites and API docs.

Caveat: any errors are our own, and not intentional.

Please email with any changes and we will update the blog post. Or, if you want any other vendors covered, let us know.

Data Connectors

First, comparing Graphit and other vendor data connectors to Carbon's offering.

We are always looking to add new data connectors upon customer request. We prioritize feature requests for paid customers on the Starter or Growth tiers.

Note, Vectara doesn't appear to have data connectors available via API, but does have an external CLI available, and we've denoted this in the table.

Costs & Data Ingestion

Second, looking at costs and ingestion capabilities.

LlamaCloud hasn't posted their pricing publicly, from what we found. LlamaParse pricing is available, but we are coming costs for RAG-ready ingestion, including vector embeddings.

Content Preparation & Retrieval

Third, looking at content preparation and retrieval capabilities.

Graphlit stands out because of its knowledge graph data model, which natively supports GraphRAG and graph retrieval. Also we support 'collections' which are grouping of content, like a Gmail label, where content can live in multiple collections.

RAG, SDKs & Other

Last, looking at RAG, SDKs and other capabilities not covered elsewhere.

This is where Graphlit shines, with its managed RAG pipeline, and access to all major LLMs and vision LLMs.

We provide image prompting, as well as robust RAG strategies for configuring all stages of the RAG pipeline.

Summary

YMMV, depending on your requirements. It is useful to look across the vendors and see what works best for you. We're biased, obviously, but these are all great products and we want our customers to make an educated choice.

Please email any questions on this article or the Graphlit Platform to questions@graphlit.com.

For more information, you can read our Graphlit Documentation, visit our marketing site, or join our Discord community.