[Feature Request] Consider a ModelFold provider preset #6942

🥰 Feature Description

Please consider a named ModelFold connection preset so its users do not have to determine the protocol-specific endpoint manually.

🧐 Proposed Solution

Review eligibility for a preset following the existing provider configuration in app/constant.ts and app/store/access.ts. Reuse the appropriate adapter; do not enable every model capability by default. The existing custom OpenAI endpoint remains the alternative. The README lists 302.AI as a sponsor; this request does not assume sponsorship or entitlement to placement.

📝 Additional Information

Submitted from ModelFold's brand account (AIModelFold), with AI assistance; this is a provider-affiliated request.

ModelFold is a third-party API aggregator. Public documentation: https://modelfold.ai/docs/ ; published chat catalog: https://modelfold.ai/docs/models/chat/ .

• OpenAI-style Chat Completions: base URL https://modelfold.ai/v1, POST /chat/completions.
• Anthropic Messages: base URL https://modelfold.ai/anthropic, POST /v1/messages.
• OpenAI Responses is a separate surface: base URL https://modelfold.ai/openai/v1, POST /responses.
• Model discovery requires the user's own ModelFold key: GET /v1/models, /anthropic/v1/models, or /openai/v1/models, respectively. The catalogs are not interchangeable. The public chat catalog includes gpt-5-2 and claude-sonnet-4-6; these are published examples, not a live authenticated model-list verification.

Scope/verification: documentation returned HTTP 200; unauthenticated GET /v1/models returned HTTP 401 as expected. No paid inference or end-to-end client test was performed for this request. Chat Completions streaming varies by model/route; please do not infer universal streaming, tools, embeddings, or full OpenAI feature parity. Any eventual preset should advertise only verified model/endpoint capabilities.