Back to list
Meituan LongCat Team Launches General 365: A Rigorous New Benchmark for AI Reasoning Evaluation
Industry NewsMeituanAI BenchmarkingReasoning

Meituan LongCat Team Launches General 365: A Rigorous New Benchmark for AI Reasoning Evaluation

The Meituan LongCat team has officially released General 365, a new benchmark designed to evaluate the reasoning capabilities of large language models (LLMs). In an initial assessment of 26 mainstream models, the benchmark revealed a significant performance gap in the industry. Gemini 3 Pro, currently regarded as one of the most advanced models, achieved a top accuracy rate of only 62.8%. More strikingly, the vast majority of the models tested failed to reach the 60% accuracy threshold, which is traditionally considered a passing grade. This release by Meituan's technical team establishes a more demanding standard for measuring AI reasoning, highlighting that current models still face substantial challenges in complex logical tasks.

美团技术团队

Key Takeaways

  • New Evaluation Standard: Meituan's LongCat team has introduced General 365, specifically designed to test the reasoning limits of AI models.
  • Industry-Wide Testing: The benchmark was applied to 26 mainstream models to provide a comprehensive overview of current AI capabilities.
  • Performance Ceiling: Gemini 3 Pro emerged as the top performer but only managed an accuracy rate of 62.8%.
  • Reasoning Deficit: Most tested models failed to achieve a 60% score, indicating a widespread struggle with the reasoning tasks presented in General 365.

In-Depth Analysis

The Introduction of General 365

The Meituan LongCat team has officially open-sourced General 365, positioning it as a new yardstick for the evaluation of artificial intelligence. Unlike traditional benchmarks that may focus on general knowledge or linguistic fluency, General 365 appears to target the core cognitive function of reasoning. By releasing this tool, the LongCat team provides the developer community with a rigorous framework to identify the strengths and weaknesses of various large language models (LLMs) in logical processing.

The decision to open-source this benchmark suggests a move toward greater transparency and standardization in how AI progress is measured. As models become more sophisticated, the industry requires more difficult and nuanced testing environments to differentiate between superficial pattern matching and genuine logical reasoning.

Benchmarking the Leaders: Gemini 3 Pro and Beyond

In the initial testing phase conducted by the LongCat team, 26 mainstream models were put to the test. The results offer a sobering look at the current state of AI development. Gemini 3 Pro, which is currently identified as the strongest model in the field, reached an accuracy of 62.8%. While this represents the leading edge of current technology, it also highlights a significant margin for improvement.

The data reveals a steep drop-off in performance beyond the top-tier models. The fact that the majority of the 26 models could not reach a 60% accuracy level—often considered the minimum standard for competency—suggests that General 365 is a highly challenging benchmark. This performance gap underscores the difficulty of the reasoning tasks included in the set and indicates that many current LLMs may still struggle when faced with complex, multi-step logical requirements.

Industry Impact

The release of General 365 is significant for the AI industry as it shifts the focus from simple performance metrics to deep reasoning capabilities. By setting a benchmark where even the most advanced models score near the 60% mark, Meituan is effectively raising the bar for what constitutes a "high-performing" model. This encourages AI researchers and developers to move beyond optimizing for existing, potentially saturated benchmarks and instead focus on the fundamental challenges of machine reasoning.

Furthermore, the benchmark serves as a reality check for the industry. While marketing for AI models often emphasizes human-like capabilities, the General 365 results demonstrate that there is still a long way to go before AI can consistently master complex reasoning tasks. This new standard will likely drive a new wave of innovation focused on cognitive depth rather than just model size or data volume.

Frequently Asked Questions

Question: What is General 365?

General 365 is a new reasoning evaluation benchmark released by Meituan's LongCat team. It is designed to provide a rigorous standard for testing the logical reasoning capabilities of large language models.

Question: How did mainstream models perform on this benchmark?

In a test of 26 mainstream models, the performance was generally low. Gemini 3 Pro led the group with a 62.8% accuracy rate, but the majority of models failed to reach a 60% score.

Question: Why is the 60% score significant in this context?

The 60% mark is often viewed as a basic passing grade or a threshold for competency. The fact that most models fell below this line indicates that General 365 is a particularly difficult test that exposes the reasoning limitations of current AI technology.

Related News

How Google Duplex Evolved to Call Businesses on Your Behalf in Gemini
Industry News

How Google Duplex Evolved to Call Businesses on Your Behalf in Gemini

A Google Research Blog entry titled “How Google Duplex evolved to also call businesses for you in Gemini” describes its subject as the evolution of Google Duplex to support calls to businesses on a user’s behalf in Gemini. The entry is dated November 20, 2026, and the supplied article text consists only of the label “Generative AI.” The headline identifies the capability and connects it to Duplex, but the available text does not explain how the system evolved, how users request a call, or where the feature is available. The headline establishes the topic, but the supplied material does not provide a technical account or rollout details.

How AI and Lean Are Helping Automate Formal Verification to Fight Software Slop
Industry News

How AI and Lean Are Helping Automate Formal Verification to Fight Software Slop

In Latent Space, Richard MacManus examines formal verification as an emerging way to fight what the headline calls “slop.” The central claim is that proving software correct with mathematics has traditionally required specialist work, and that AI and Lean are now helping automate that process. The distinction is important: the reported change concerns how the work of proving correctness is done, not what counts as a proof. The supplied excerpt presents automation as assistance, without saying that it replaces specialists or completes proofs independently. It also does not specify which parts of the process AI and Lean help automate, how much work they save, or which software has been verified this way.

AI and Human Claim Solutions to the 167-Year-Old Riemann Hypothesis
Industry News

AI and Human Claim Solutions to the 167-Year-Old Riemann Hypothesis

TechCrunch AI reports that AI claimed a solution to the Riemann hypothesis, a mathematical problem the headline describes as 167 years old. A human also put forward a competing solution. The report, written by Tim Fernholz and published on November 20, 2026, describes mathematicians as astonished by the AI-driven result. The supplied excerpt does not explain either proposed solution, identify the AI system or the human claimant, or say whether either claim has been verified. The distinction between a claimed solution and a confirmed one is central to reading the report: the excerpt establishes that two claims are being reported, but it does not establish that either has resolved the hypothesis.