Unverified80% confidenceFactTime unknown
MTP-LX 4bit版本在理财规划推理任务上经GPT-5.5 Thinking评分为55分（满分100），而同题GPT-5.5 Pro得分82分
1
Sources
80%
Confidence
Medium-term (~90 days)
Relevance
5/31/2026
First Seen
Valid until: 8/29/2026
Sources
Mac本地跑Qwen3.6-27B：4种方案实测对比
bilibilikate人不错
Related Entities
Related Claims
Unverified顶级自回归模型如GPT-4o、Claude 3.5在GSM8K上准确率已超过95%，导致基准饱和71% similarVerifiedGPT-5.4在SWE-Bench Pro测试中拿下57.7%准确率，超越了GPT-5.3-Codex的56.8%71% similarUnverified在max模式下GPT-5.6的得分反而略低于ultra，印证了过度思考现象71% similarUnverifiedGPT-5.6编程能力在关键基准上达到91.9%，安全评测CyberGym得分85.6%，均超越竞品Nexus 570% similarUnverified在Agents' Last Exam评测中，GPT-5.6 Sol取得53.6分，比采用自适应推理的Claude Fable 5高出13.1分70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/4337API
curl https://kongchang.com/api/v1/knowledge/claims/4337MCP
get_claim(id=4337)