[enhancement] support more LLM backends

[shaoyuyoung]

In recent months, DeepSeek API has been really annoying😡, which always gets interrupted for no reason.
Even though I added a retry mechanism (#5), it was of no avail.
We must support more powerful and stable LLMs (e.g., GPT-4o, Claude-4, Gemini-2.5, Kimi, Qwen).

• deepseek-v3
• kimi-k2
• gpt-4o
• gemini-2.5

[shaoyuyoung]

cc @lingebeng , take a look at this issue: MoonshotAI/Kimi-K2#9.

Not sure whether we can run this example on kimi-k2.

Previously, in my contributed code, the stop param always throws ConnectionError in FUEL.

[shaoyuyoung]

cc @lingebeng , take a look at this issue: MoonshotAI/Kimi-K2#9.

Not sure whether we can run this example on kimi-k2.

Previously, in my contributed code, the stop param always throws ConnectionError in FUEL.

This has been addressed