Benchmark 2.0

[Meldiron]

🔖 Enhancement description

Details to include:

• ✅ Tokens used for each question (and average per model)
• ✅ Separate model input & output price
• ✅ Duration for each question (and average per model)
• ✅ Cost for each question (and average per model)
• ✅ TPS of model, like price of model
• ✅ total cost (sum of all)
• Structure that stores old benchmarks too, not just latest
• Remember total tool calls done for each question, and average on model

🎤 Pitch

More insightful benhmark

👀 Have you spent some time to check if this issue has been raised before?

• I checked and didn't find similar issue

🏢 Have you read the Code of Conduct?

• I have read the Code of Conduct