Unverified50% confidenceBenchmarkExact time
On the public KIA Soul dataset, TEMPEST outperforms the best classical model by 7.3 percentage points within-session and by 14.3 percentage points cross-session.
1
Sources
50%
Confidence
Long-term
Relevance
10/7/2026
First Seen
Sources
Related Entities
Related Claims
Partially VerifiedIn the KingBench 3 evaluation, Soul scored 55 (approximately 78.6%), Tara scored 44 (approximately 62.9%), and Luna scored 31 (approximately 44.3%).56% similarUnverifiedQwen3.7 Max的输出质量高度依赖提示词质量，同一模型在不同提示词下的性能差异可高达30%-50%53% similarUnverifiedIn Theo's tests, GPT-6.1 Sol reads an average of only about 110,000 tokens per request, whereas Sonic 5 reads as many as 360,000 tokens.52% similarUnverifiedOn the Deep Squeeze test, the old model scored 49%, while Gemini 3.7 Flash jumped to 65.3%.51% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/983179API
curl https://kongchang.com/api/v1/knowledge/claims/983179MCP
get_claim(id=983179)