Unverified50% confidenceTradeoffExact time
Although full-modal large models support voice input and output, their scores on agent benchmarks are far lower than those of mainstream reasoning models.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/31/2026
First Seen
Valid until: 11/29/2026
Sources
Related Entities
Related Claims
Unverified非自回归语音模型通常比自回归模型快10-100倍，但表现力受限于预测器准确性79% similarUnverifiedMultiple independent studies in 2023 showed that some top models scored extremely high on standard mathematical reasoning tests but performed significantly worse on equivalent rephrased versions, indicating that the models learned test patterns rather than genuine reasoning ability.72% similarUnverified在事实性评估任务上模型评判者准确率仍显著低于人类专家，但在风格、连贯性等主观维度已接近人类评审员间一致性水平70% similarUnverifiedResearch shows that a model's accuracy on self-criticism tasks is significantly lower than in scenarios where criticism is performed by another model.70% similarUnverifiedMultimodal large models have extremely low factual recall accuracy in video question answering, which is difficult to improve even with supervised fine-tuning.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/830624API
curl https://kongchang.com/api/v1/knowledge/claims/830624MCP
get_claim(id=830624)