Dissatisfied with performance: Meta allegedly postpones release of AI model
Meta originally planned to release its next AI model, "Avocado," in March. Since its performance was not convincing, it is now reportedly not expected until May
(Image: JarTee/Shutterstock.com)
Meta is dissatisfied with the performance of its next AI model, which is why the technology codenamed Avocado is to be released two months later than planned. This is reported by the New York Times, citing internal sources. The AI model was originally supposed to be released in March, but a release is now planned for May, according to the report. The reason for the postponement is said to be the insufficient performance compared to the current AI technology of competitors Google, OpenAI, and Anthropic in the areas of reasoning, coding, and writing. Internally, there was even discussion about temporarily licensing Google's AI Gemini for Meta's AI products. However, no decision has been made yet.
Meta remains confident
A postponement of the AI model would be a setback for Meta. The Facebook group intends to spend over 115 billion US dollars this year on the further development of its own AI technology. Nevertheless, its competitors' technology is currently considered superior, the US newspaper further explains. "As we've said publicly, our next model will be good. But more importantly, it underscores the rapid trajectory we're on," the company has now assured Reuters and: "We're excited for people to see what we've been cooking very soon." The US company has not yet commented on the postponement of the release.
Videos by heise
At the beginning of February, it was reported that the pre-training of Avocado had been successfully completed. It surpassed the best freely available base models and was able to keep up with leading pre-trained models. At the same time, it is said to have been significantly more efficient than previous Meta models. The dissatisfaction may therefore have arisen only afterwards. Last year, Meta already had considerable problems with the AI model Llama 4. The release was postponed several times, the company manipulated benchmarks, and developers were disappointed with the performance. Meta subsequently restructured its AI department.
(mho)
