Unverified50% confidenceBenchmarkExact time
In an evaluation of 30 answerable questions, off-the-shelf small models under 350MB achieved faithfulness (all assertions can be found in the passages) at most 3 times.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn a small-scale test published by the PartsPlan developer, all 30 chart values were answered correctly.68% similarUnverifiedResearch shows that 30B-parameter models have a success rate 15–30 percentage points lower than GPT-4/Claude-3.5 on complex reasoning tasks.66% similarUnverified千问三235B MoE总参数量2350亿，但每次推理实际激活的参数约22B65% similarUnverifiedA 70B-parameter model requires approximately 35–40GB of VRAM after Q4 quantization, which a single T4 cannot handle, but dual T4s can basically satisfy.64% similarUnverified通义千问3.6 27B在语义三次全对率和格式正确率上均领先Muse Glimmer64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1013587API
curl https://kongchang.com/api/v1/knowledge/claims/1013587MCP
get_claim(id=1013587)