feat: LLM routing integration via NotDiamond

[barabasicsongor]

Overview

Hi there Langflow team! I am a big fan of Langflow and would love to integrate Not Diamond into it. In case you're unfamiliar with Not Diamond, it automatically routes your prompts to the AI model best-suited to respond to that prompt. By using Not Diamond as an AI "meta-model," you can achieve state-of-the-art performance that beats every foundation model on every major benchmark.*

For Langflow we would like to integrate our router into your app as a component, so that users can automatically route their prompts across multiple models.

Description of approach

We have integrated NotDiamond as a new component under the Prototypes group. The call to NotDiamond is happening as a POST request API call. We have decided to go down this route, as integrating our Python SDK caused conflicts - since both ours and your library is inheriting from Langchain.

Users can choose the Not Diamond Router component, it accepts an input text/message/prompt, and the output will be a string that represents the model they should call next.

This output will be integrating perfectly with your conditional component soon.

Screenshots:

Other relevant notes

*If you’d like to learn more about our routing, check out this example walkthrough video with Tomas, our CEO. You can also
see how we work by using our Chat app,
learn more at our website, or
read our docs.

[codspeed-hq]

CodSpeed Performance Report

Merging #4543 will degrade performances by 28.56%

_{Comparing Not-Diamond:nd-integration (9c0e8a8) with main (ab71e2f)}

Summary

❌ 1 regressions
✅ 14 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	Not-Diamond:nd-integration	Change
❌	test_successful_run_with_input_type_text	380 ms	531.9 ms	-28.56%

[ogabrielluiz]

Hey @barabasicsongor

This is looking good. Thanks!

What do you think about using a HandleInput with is_list=True and allow the User to pass the LLM instances into NotDiamondComponent and using a function like get_model_name (https://github.com/langflow-ai/langflow/blob/main/src/backend/base/langflow/base/models/model_utils.py#L1) you could get the names, validate them and even call the LLM by passing it to get_chat_result(https://github.com/langflow-ai/langflow/blob/main/src/backend/base/langflow/base/models/chat_result.py#L40).

[barabasicsongor]

Hi @ogabrielluiz , that's is a great suggestion! I've just updated the PR, if you could have a look please 😄
I've also extended the supported Anthropic models list as they were missing the latest versions.

There is 1 minor thing which I haven't figured out yet, so if you have any suggestion would be amazing:
How to access the stream value from previous nodes in my Not Diamond node, so I can pass it on? Currently I only get access to the Langchain Runnable models, which don't contain that.

[Cristhianzl]

Regarding the FE side - new bundle added.

LGTM