Skip to content
IFAI

AI models

OpenAI's updated GPT-4o reclaims the crown on Chatbot Arena

Published Reach 51

OpenAI just took back the #1 spot on the LMSYS Chatbot Arena leaderboard with its updated ChatGPT-4o model, edging past Google's Gemini-1.5-Pro-Exp with a 1314 score. Tested anonymously across over 11,000 community votes, the new checkpoint shows noticeable gains in coding (up 30+ points) and hard prompts, and developers can already call it in the API.

OpenAI's updated GPT-4o reclaims the crown on Chatbot Arena
Image: x.com

The heavyweight title in AI just changed hands once again.

LMSYS announced that OpenAI's latest ChatGPT-4o checkpoint (20240808) has reclaimed the #1 spot on the Chatbot Arena leaderboard with a score of 1314, nudging past Google's experimental Gemini-1.5-Pro-Exp after more than 11,000 blind community votes.

Coding gains and a fresh API endpoint

The biggest leap came in technical areas. In coding, the new checkpoint scored over 30 points higher than the original May version of GPT-4o. It also swept top rankings across hard prompts, instruction-following, longer queries, and multi-turn conversations, while tying for first in math.

OpenAI made the checkpoint immediately available in its API under the name chatgpt-4o-latest. The team noted that gpt-4o-2024-08-06 remains the recommended choice for standard production workloads, while the latest endpoint is aimed at developers looking to evaluate cutting-edge chat improvements.

Why it matters

We are watching the frontier model race turn into a weekly slugfest. Google barely had time to celebrate its experimental Gemini win before OpenAI pushed out a tuned checkpoint to reclaim the crown.

Splitting the release into a rolling chatgpt-4o-latest endpoint alongside stable dated snapshots is a smart move. It allows OpenAI to show off its absolute peak capabilities on community benchmarks without breaking production pipelines for developers who need predictability. If your workflow relies heavily on code generation or complex instructions, it is definitely worth running a few test prompts against the latest endpoint today.

Sources (1)

  1. [1]
    @lmsysorg(opens in a new tab)Originalx.com

Topic trackerOpenAI GPT-4o model updates

Every story on this topic, newest first.

  1. OpenAI updated GPT-4o — and can't quite explain why it's better
  2. OpenAI's updated GPT-4o reclaims the crown on Chatbot ArenaThis story
  3. OpenAI quietly updates GPT-4o in ChatGPT

Similar events

Past happenings like this one, for context.

Best of AI, daily