Next-Gen AI Models Race Heats Up: GPT-5.6 and Qwen3.8-Max Anticipated

The competition for advanced AI capabilities intensifies with OpenAI's GPT-5.6 expected for general API availability in August-September, and Alibaba's Qwen3.8-Max promising open weights and general availability soon, pushing the boundaries of AI applications.

The top 3

  1. Highest MMLU Scores Among Leading LLMs: OpenAI's GPT-5 leads the MMLU benchmark with a score of 92.5%, followed closely by Anthropic's Claude Opus 4.6 at 92.1% and OpenAI's o1 at 91.8%.
  2. Largest Parameter Counts of Advanced LLMs: Google AI's PaLM model boasts 540 billion parameters, while Meta's Llama 3.1 405B features 405 billion parameters, and DeepSeek-V3 has 671 billion parameters, showcasing the scale of modern LLMs.
  3. Most Adopted Commercial and Open-Source LLMs: OpenAI's GPT-4o is recognized as one of the most advanced and widely used commercial models, while Meta's Llama 3.1 is a leading open-source model, and Google Gemini is a versatile top contender.

Sources

Open the full topic