Add AnyLLM bot — one API key brings 24+ Chinese & international LLMs into ChatALL

Hi team 👋

I've been using ChatALL as my go-to tool for comparing LLM answers side by side — the concurrent multi-bot design is exactly what's missing from most chat UIs.

Looking at the open issues, I see a recurring pattern: requests to add DeepSeek (#967, open since Jan 2025), latest Claude/Gemini models (#1017, #1025), and general "please update to all recent models" (#946, #964). Each new model today means a separate bot integration with its own auth flow — that's a heavy maintenance burden, and it's why the backlog keeps growing.

I'd like to propose a different approach: an AnyLLM bot that's API-based (not web-login-based). AnyLLM is a unified OpenAI-compatible API gateway — one API key gives access to 24+ leading LLMs:

Chinese: DeepSeek V4 Flash/Pro, Qwen3.7-Plus, GLM-5.2/5.1 (up to 2M context), Kimi K2.5/K2.6, MiniMax M3/M2.7
International: GPT-5.6 Luna, GPT-5.4 Mini/Nano, Gemini 3.5 Flash (multimodal)
One bot class, one auth field (API key), one endpoint — but users get 24+ models they can select between. This means:

DeepSeek support lands immediately (addresses #967) without a separate DeepSeek bot
New models require zero code changes in ChatALL — they appear automatically as AnyLLM adds them to the gateway
International users get Chinese models without Chinese phone numbers / Alipay (a known friction point)
ChatALL's compare-answers UX gets more candidates per query — users can fire a prompt to DeepSeek + GPT-5.6 Luna + Gemini 3.5 Flash simultaneously through one bot
I've read the "How to Add a New AI Chatbot" wiki and CONTRIBUTION.md. I'm comfortable implementing the bot class following the existing API-based bot pattern (similar to how OpenAI API bot works), with Prettier formatting and passing the build.

Before opening a PR, a few questions:

1.Is an API-based "multi-model gateway" bot welcome, or do you prefer each model as a separate bot for UI granularity (individual on/off toggles per model)?
2.For model selection within the bot — should each model be a separate "bot instance" the user enables independently, or a single bot with a model dropdown in settings?
3.Any preference on bot ID / naming? I'd use AnyLLM to match the gateway branding.
Happy to submit the PR with tests, maintain it long-term, and keep the model list in sync as the gateway updates.

Thanks for building ChatALL — the "discover the best answer by asking everyone at once" concept is still underrated.

— Leo Bennett [leo.indiedev@gmail.com]