GPT-4 Turbo Update! Testers Say: It’s Even Smarter

April 9, 2024 – Recently, OpenAI officially announced that the new version GPT-4 Turbo is now available to all paying ChatGPT users. If your knowledge cutoff date has been updated to April 2024, then you have successfully received the new version update. If not, please be patient for a while longer.

Enhanced Capabilities and User Experience

According to reports, the new version, with its “less words, more action” approach, has seen significant improvements in various areas such as writing, mathematics, logical reasoning, and coding.

When using ChatGPT for writing, you’ll notice faster responses, more direct communication, and a greater use of colloquial expressions.

In essence, the new version is more aligned with human natural language habits, adding a touch of humanity while reducing some of the AI flavor.

For instance, when sending a text to remind a friend about a birthday dinner invitation, the previous versions might have been verbose, akin to a mini-Shakespeare, whereas the current responses are more concise, conveying the core message directly.

new gpt4 turbo chat test

Development and Rollout

In fact, the debut of GPT-4 Turbo can be traced back to last November’s OpenAI Developer Conference, but it was initially provided to developers as a “preview version.” It’s been almost five months now, and OpenAI has officially rolled out this new version via API.

Here’s some parameter information:

  • Fully available, accessible via “gpt-4-turbo” to use the new model.
  • Major upgrade in basic capabilities.
  • Comes with image reading ability, no need for 4v interface.
  • 128K context.
  • Enhanced coding abilities.

Shortly after its release, the Poe platform swiftly integrated it into its services. If you’re a Poe member, you’re likely already enjoying the benefits.

OpenAI CEO Greg Brockman promptly shared users’ test results on social media to promote the new product.

gpt4 turbo users test results

Benchmark test results indicate comprehensive improvements compared to last November’s version. Although OpenAI is still a bit tight-lipped, the user experience may have improved significantly.

chatgpt4 Benchmark test results

Competition and Rankings

Claude 3 Opus emerged victorious in the arena back in March, surpassing GPT-4 to claim the top spot. OpenAI, preoccupied with debates about human welfare with Musk, remained silent when everyone was anticipating the release of GPT-4.5/5.

musk suing openai

It wasn’t until Gemini announced the official release of Gemini 1.5 Pro that OpenAI swiftly responded by providing the new version GPT-4 Turbo to paying users via API.

Following the announcement, the new version was promptly implemented. OpenAI CEO Sam Altman also hyped the new version of GPT-4 on social media, stating that “GPT-4 is now noticeably smarter and more enjoyable to use.”

Regarding this update, reliable insider @futuristflower revealed some insider information.

They’ve got something big – let’s call it GPT-5 – but they’re holding back. So, they trained GPT-4 against it, which squeezed out a few improvements, but not enough to call it 4.5 (told you so months ago!). Now they’re stuck. No amount of finetuning will make GPT-4 any better now.

According to The Information, two OpenAI researchers were dismissed for allegedly leaking confidential information, including Leopold Aschenbrenner, a loyal ally of last year’s palace intrigue protagonist Ilya Sutskever.

Another dismissed researcher, Pavel Izmailov, was also a member of the security team. Although it’s unclear what specific information these two dismissed employees leaked, this news is intriguing at this pivotal moment of OpenAI’s accelerated progress.

It’s worth noting that more than 8,000 users have once again voted to propel the new version GPT-4 back to the top of the leaderboard, returning to its “familiar” position.

8k user vote results

The model’s performance varies across different languages. GPT-4 Turbo still maintains a significant lead in English proficiency, but in other languages, such as Chinese, it appears somewhat lacking, slightly behind Claude-3 Opus.

turbo model language performance tests

Test results can be found here.

Just as people were immersed in exploring Gemini 1.5 Pro, OpenAI’s announcement of the official release of GPT-4 Turbo with major upgrades and image reading functionality has once again shifted the spotlight.

Gemini 1.5 Pro, which should have been a Google powerhouse, has become yesterday’s news. With this latest release of GPT-4 Turbo integrated into ChatGPT, OpenAI not only exerts pressure on Claude 3 but also steals the limelight from Google once again.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top