OpenAI has made the GPT-4 Turbo with Vision generally available in the API as well as in ChatGPT. The company announced the update on ‘X’ today saying the model has been “greatly improved” without specifying what these changes were. The new version will also have the knowledge cutoff date till December 2023.
The model will now have data processing capabilities to analyse text in the JSON format and perform function calling which generates a JSON code snippet for developers to automate actions within their connected apps like sending mails, posting online or making purchases.
Users on ‘X’ responded to the announcement with curiosity around whether improvements had been made around speed or accuracy.
Owen Campbell-Moore, a Product Manager with an AI firm tweeted the announcement saying the model was a “mega improvement over previous ones, especially at math” suggesting that the visual analysis function wasn’t the only major addition. He replied to a request for more information saying, “I’m hoping we can get evals out shortly to help quantify this. Until then - it’s a new model with various data and training improvements, resulting in better reasoning.”
Earlier in January this year, the Sam Altman-led company released several updates to its AI models and pricing to tackle increasing competition. The pricing for GPT-3.5 Turbo was lowered along with a performance boost given to GPT-4 Turbo after user complaints around the model’s “laziness.”
As the industry awaits the release of GPT-5, other models have gradually caught up with the capabilities of the predecessor. Anthropic’s Claude 3 Opus released in March has beaten GPT-4 on the leaderboard for closed-source models with Google’s Gemini Pro close on the heels.
