Expired50% confidenceBenchmarkExact time
According to its blog data, Longcat 2.0 outperformed several closed-source flagship models on SWE-Bench Pro.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/8/2026
First Seen
Valid until: 10/6/2026(expired)
Sources
Related Claims
UnverifiedLongcat 2.0 is an MoE model with 1.6 trillion parameters, activating only approximately 48 billion parameters at a time, and is natively designed for agentic scenarios.66% similarUnverified美团的大模型Longcat 2.0据称达到1.8T参数，规模超过DeepSeek旗舰产品66% similarUnverifiedRoPE and its improved variants YaRN and LongRoPE have become the mainstream positional encoding choices for current ultra-long-context models.57% similarUnverifiedGLM 5.2 relies on extremely long chains of thought, with token output that even exceeds that of DeepSeek version 2.6.54% similarUnverifiedThe DeepSWE benchmark was introduced by datacurve.ai and is described as a 'long-cycle software engineering benchmark'.53% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/291498API
curl https://kongchang.com/api/v1/knowledge/claims/291498MCP
get_claim(id=291498)