Qwen3 Max's main index dropped 10.9 points in today's Smoke test, with the code execution dimension falling from a perfect 100 to 75. This one-day fluctuation exceeds the normal random variance and requires serious attention.
Today's Smoke lightweight evaluation results show Doubao Pro leading with 97.75 points (Execution 100, Constraint 95), becoming the only model among 11 mainstream models to break 97 points on the main ranking. GPT-5.5, which was previously expected to perform well, scored only 60.58, dropping 23.5 points compared to yesterday.
In this WDCD cycle, GPT-5.5 re-establishes the ceiling of instruction adherence with an absolute score of 71.67, while Gemini 2.5 Pro's 14.2-point leap completely overturns the perception that Google models are weak in adherence. Meanwhile, Wenxin Yiyan 4.5 suffers a 7.5-point drop, signaling potential over-alignment issues.
The WDCD five-scenario evaluation reveals that resource constraints is the hardest scenario with the lowest overall scores, while DoubaoPro achieves the highest score in business rules, demonstrating significant model specialization.
The WDCD three-round test reveals that model integrity drops to 30.6% under direct pressure in R3, with Grok4 hitting a 93.3% collapse rate, exposing the fragility of safety alignment.
The WDCD Commitment Test reveals models' true performance under constraints through three rounds of dialogue. GPT-5.5 leads with 71.67 points, while Grok 4 scores only 52.5 points, ranking last—a gap of 19.17 points between the top and bottom.
Claude Sonnet 4.6 showed abnormal results in today's Smoke test, with the material constraint dimension dropping sharply. The drop may be due to sampling variance but warrants further monitoring.
In today's Smoke evaluation, Claude Opus 4.7's main score dropped by 9 points from 97.75 to 88.75, primarily due to a sharp decline in the material constraint dimension from 95 to 75 points—a direct loss of 20 points in a single day.
This week's 7-day Smoke Quick Test data reveals polarization: Wenxin Yiyan surged 53.4 points while GPT-o3 fell 7.8 points.
Today's Smoke Lite evaluation results show a three-way tie for first place at 88.75 points, while the Claude series suffered sharp declines. The shakeup signals that open models are rapidly closing the gap with closed-source leaders.
GPT-5.5's code execution score dropped from 100 to 50, causing a 28-point drop in the main ranking. But is this degradation or just sampling noise?
Gemini 2.5 Pro's credibility rating fell from pass to fail, causing a 10-point drop in the main ranking, even though its code execution score remained perfect.
Today's YZ Index Smoke lightweight test reveals that three leading models suffered significant drops, while Claude models dominate near-perfect scores with structural advantages in code execution and material constraint.
Today's Smoke evaluation shows the main benchmark up by 5 points, but the integrity rating drops from pass to fail, signaling a classic alarm of "seemingly stronger capability but lost trustworthiness at the admission gate."
In today's Smoke evaluation, Anthropic's Claude Sonnet 4.6 saw a dramatic split: material grounding scores dropped 27.5 points to 69, while code execution surged 25 points to a perfect 100, with the main leaderboard edging up 1.4 points to 86.05.
Today’s Smoke benchmark shows Claude Opus 4.7 leading with 88.75, while two models scored zero in code execution; the real differentiator is material constraint, not execution ability.
GPT-OSS 20B is a new Mixture-of-Experts (MoE) pretraining benchmark introduced in MLPerf Training v6.0, designed to lower the barrier to entry and evaluate sparse architectures. It features variance reduction techniques to ensure fair comparisons and can run on a single 8-GPU node.
In today's Smoke Evaluation, Claude Opus 4.7's main chart score plummeted from 89.43 to 79.86, a net loss of 9.6 points, with code execution collapsing from a perfect 100 to 75. The sharp drop raises the question of whether this signals model degradation or is merely a random sampling fluctuation.
In today's Smoke evaluation, Claude Sonnet 4.6's code execution score dropped from a perfect 100 to 75, directly dragging down the main leaderboard score by 4.2 points. This is not a minor fluctuation but a potential signal: is the model truly degrading, or is it the randomness of daily sampling at play?
In the Smoke Lite evaluation on May 14, 2026, the key finding is shocking: Claude Sonnet 4.6 surged to the top with a main score of 84.68, but the code execution dimension of 8 mainstream AI models collectively dropped by 25 points, causing a drastic reshuffle in overall rankings. This is no coincidence—it’s a hidden crisis signal of rapid iteration in the AI industry.
