Re:GPT-4.5 or GPT-5 being tested on LMSYS?
by monday40 on Monday April 29, 2024 @08:50PM
(#40206097)
Hmm... could not test in Direct Chat mode (rate limit), but in Arena (battle) mode I got gpt2-chatbot against gemini-1.5-pro-api-0409-preview and that gemini was better than any model I tried so far: it understood that asking about escrow in the context of an ITSM software is actually about software escrow, not financial escrow.
The prompt was: Does HaloITSM have escrow functionality?
The prompt was: Does HaloITSM have escrow functionality?