Unverified50% confidenceBenchmarkExact time
On AdvBench against DeepSeek-V4-Flash, SEMA-Auto achieves 76.0% ASR@1 within an average of 3.0 turns.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/27/2026
First Seen
Valid until: 1/25/2027
Sources
Related Entities
Related Claims
UnverifiedOn AdvBench against DeepSeek-V4-Flash, the best static baseline achieves only 54.8% ASR@1 with a fixed 7 turns.72% similarUnverifiedDeepSeek's model V4 Flash 0731 achieved a score of 82.7% on Terminal-Bench 2.1.59% similarUnverifiedWith a success rate comparable to DeepSeek V4 Pro, the per-task cost of GLM-5.3 Flash is only one-eighth that of V4 Pro.59% similarUnverifiedIn the receipt recognition task, DeepSeek V4.1 Flash was twice as fast as Haiku 5.5.58% similarUnverifiedz-ai/glm-5.3 has a context length of 1M and ranks second on the AA open-source index; z-ai/glm-5.3-flash ranks fourth; deepseek-ai/deepseek-v4.1-flash ranks fifth; moonshotai/kimi-k2.6 has a context of 262K.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1103174API
curl https://kongchang.com/api/v1/knowledge/claims/1103174MCP
get_claim(id=1103174)