Document Requesty as a supported LiteLLM provider

[Thibaultjaigu]

Documents Requesty as a supported provider via LiteLLM's requesty/ model prefix, mirroring how OpenRouter is presented.

Requesty (https://requesty.ai) is an OpenAI-compatible LLM router that LiteLLM natively routes via the requesty/ prefix (auto-reading REQUESTY_API_KEY). markpdfdown's LLMClient passes the model name straight to LiteLLM with no provider allowlist, so no code change is needed, just config/docs.

Changes:

• src/markpdfdown/config.py: the MODEL description now includes a requesty/openai/gpt-4o example.
• README.md: Requesty in the multi-provider line, a REQUESTY_API_KEY env example, a "Requesty Models" section (MODEL_NAME=requesty/openai/gpt-4o, requesty/anthropic/claude-sonnet-4-5, verified live), and a Docker example. Key at https://app.requesty.ai/api-keys.
• .env.sample: Requesty API key block + usage example.

config.py parses cleanly.

I work at Requesty. This mirrors the existing OpenRouter provider as closely as possible. Happy to adjust or close it if it's not a fit.

[jorben]

Glad to receive Requesty’s PR. I’ll merge it.

If you’re interested, you’re also welcome to submit Requesty as a provider to my latest LLM gateway project, TiyGate. You can refer to the existing provider implementations for OpenRouter and ZenMux.

[Thibaultjaigu]

Hi @jorben, thanks for the kind words back in July. Just checking in on this one, it is still open and rebases cleanly on master. Let me know if you want anything changed before merging.

And thanks for the TiyGate pointer, I will take a look at the OpenRouter and ZenMux providers there and open a PR for Requesty.