Unverified50% confidenceFactExact time
ECMWF从2016年开始制作ERA5，利用IFS模型CY41R2版本重新模拟，总数据量超过5PB
1
Sources
50%
Confidence
Long-term
Relevance
8/11/2026
First Seen
Sources
DeepMind开源WeatherNext：AI气象预测新标杆
Related Claims
UnverifiedECMWF's ERA5 uses a four-dimensional variational assimilation (4D-Var) method, performing one assimilation cycle every 12 hours and fusing approximately 40 million observational data points.71% similarUnverifiedLFM2.5-8B-A1B在工具调用（tool calling）场景中官方声称可达到4倍体量模型的效果62% similarUnverifiedOmniMerge V4 is a model formed by fusing three to four 27B models through weight merging.56% similarUnverifiedThe base model (no LoRA) was configured with 30 steps, CFG 4, er_sde sampler, and simple scheduler56% similarUnverifiedThe Spark large model has been iterated through multiple versions and has developed to version 4.0.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/731087API
curl https://kongchang.com/api/v1/knowledge/claims/731087MCP
get_claim(id=731087)