Unverified50% confidenceBenchmarkExact time
On LongBench v1/v2 and MemoryAgentBench, NoteBank also reported improvements, covering everything from three 1.3B research models up to Mamba-2-8B and a trained Qwen3.5-9B.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/15/2026
First Seen
Valid until: 1/13/2027
Sources
Related Entities
Related Claims
UnverifiedQwen3 27B's performance on coding and agent benchmarks is: SWE-bench Pro 61.7, Terminal-Bench 2.1 73.0, and LiveCodeBench V6 90.3.63% similarUnverifiedIn the experiment, the Qwen3.5-0.8B backbone and approximately 51 billion parameter PLE memory module were frozen, with only R=1 readers deployed at decoder layers 3 and 9 being trained.61% similarUnverifiedThe experiments covered three different model families: Qwen14B, Qwen7B, and DeepSeek6.7B.60% similarUnverifiedAdaptiveSpec supports three mainstream target models: DeepSeek-R1-Distill-Llama-8B, Llama-3.1-8B-Instruct, and Qwen3-8B.59% similarUnverifiedMemory frameworks such as Mem0 and Zep are exploring standardized long-term memory interfaces for AI.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1040237API
curl https://kongchang.com/api/v1/knowledge/claims/1040237MCP
get_claim(id=1040237)