BFCL May 14th Release (GPT-4o and Gemini)

[Fanjia-Yan]

This PR makes 3 models(4 entries) available for inference on BFCL:

• gpt-4o-2024-05-13 (Function Calling Mode and Prompting Mode)
• gemini-1.5-pro-preview-0514 (Function Calling Mode)
• gemini-1.5-flash-preview-0514(Function Calling Mode)

You can start the evaluation by running python openfunctions_evaluation.py --model MODEL_NAME and get score by running python ./eval_runner.py --model MODEL_NAME. For more detail, refer to Readme under the BFCL page.

Score changes are reflected in #428 .

This PR also updated different models' pricing:

• For Gemini, when prompts are less than 128K tokens, the new Gemini series' prices are lowered by around half(https://ai.google.dev/pricing). All the BFCL test cases are less than 128K tokens.
• For Anthropic Models, the prices have decreased for claude-2.1 and Claude-instant-1.2 which have updated accordingly
• For Mistral Models, the prices have been halved for Mistral-large and Mistral-Small
• For OpenAI Models, we have corrected GPT-3.5-turbo-0125 to the price it should have

[CharlieJCJ]

Reviewed costs and synced the most up-to-date costs with various model providers.

[HuanzhiMao]

Ready to ship