iPhone 17 Air leaked images.

Meta Under Fire for Manipulating Llama 4 Benchmark, But It Isn’t the First Time

Meta's latest Llama 4 release faces scrutiny amidst mixed user reviews.

On 1Point3Acres, a popular forum for Chinese people in North America, a user claiming to be a former Meta employee posted a bombshell. According to the post, which has been translated into English on Reddit, the Meta leadership allegedly mixed “the test sets of various benchmarks in the post-training process” to inflate the benchmark score and meet internal targets.

Meta's GenAI Head , Ahmad Al-Dahle , strongly denies training on test sets, attributing variability to implementation stabilization.

However, this isn't unprecedented. Research previously showed significant benchmark data contamination in Llama 1's pre-training corpus.

Adding to concerns, LMSys Arena noted Meta lacked clarity about an experimental "Llama-4-Maverick" model, prompting leaderboard policy updates for transparency.

iPhone 17 Air and iPhone 16e in comparison

Snapdragon 8s Gen4 is more efficient than Snapdragon 8Gen3 and Dimensity 8400 Ultra

#Qualcomm #Snapdragon8sGen4