{"success":true,"data":{"author":"Google Research (NeuralGCM team)","bucketAccessType":"public","bucketListable":true,"bucketName":"neuralgcm","bucketPrefix":"models/v1","category":"Machine Learning & AI","cloudProvider":"gcp","createdAt":"2026-11-14T14:15:26.024Z","description":"Trained weights for NeuralGCM, the hybrid model of Kochkov et al. (Nature, 2024) that couples a differentiable dynamical core for atmospheric dynamics with learned neural network physics, trained on ERA5 to produce both medium-range weather forecasts and multi-year climate simulations. Measured 2026-11-14: this folder (models/v1) holds deterministic checkpoints at 0.7, 1.4 and 2.8 degree resolution and a stochastic 1.4 degree checkpoint, 4 pickle files totalling about 0.31 GB; the sibling folder gs://neuralgcm/models/v1_precip holds two stochastic 2.8 degree checkpoints that predict precipitation and evaporation (one trained on precipitation, one on evaporation). The NeuralGCM documentation recommends the 1.4 degree stochastic model for weather forecasts, the 0.7 degree deterministic model for the first few forecast days, and the 2.8 degree stochastic models for multi-decadal atmosphere-only climate runs, which need external sea surface temperature and sea ice forcing. Load a checkpoint with pickle and pass it to neuralgcm.PressureLevelModel.from_checkpoint (JAX), then initialise from ERA5 such as the ARCO-ERA5 Zarr. Checkpoints are distributed by Google under CC BY-SA 4.0; the code is Apache 2.0.","displayTitle":"NeuralGCM Pre-trained Model Checkpoints (v1)","doi":"10.1038/s41586-024-07744-y","fuseMountBrowsable":true,"id":"3410c175-c86c-57e5-a550-33dac91e35bf","license":"CC-BY-SA-4.0","licenseEvidenceUrl":"https://github.com/neuralgcm/neuralgcm/blob/main/docs/checkpoints.md","probeStatus":"ok","referenceUrl":"https://doi.org/10.1038/s41586-024-07744-y","region":"us","schemaCard":{"accessHints":[{"language":"python-gcsfs","snippet":"import gcsfs\nfs = gcsfs.GCSFileSystem(token='anon')\nfs.ls('neuralgcm/models/v1/', detail=False)[:10]"},{"language":"python-storage","snippet":"from google.cloud import storage\nclient = storage.Client.create_anonymous_client()\nlist(client.bucket('neuralgcm').list_blobs(prefix='models/v1/', max_results=10))"}],"caveats":["Bucket has a flat key structure (no prefix folders). Avoid unbounded LIST — page with pageToken or filter by prefix.","Files are binary — use a domain-specific parser rather than text I/O."],"format":"binary","generatedAt":"2026-11-14T14:15:30.933Z","layout":{"example":"models/v1/","pattern":"{id}/{seg1}/{object}"},"mountPath":"/mnt/gcs-archive/data","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"models/v1/deterministic_0_7_deg.pkl","truncatedAt":512}],"sampleObjects":["models/v1/","models/v1/deterministic_0_7_deg.pkl","models/v1/deterministic_1_4_deg.pkl","models/v1/deterministic_2_8_deg.pkl","models/v1/stochastic_1_4_deg.pkl"],"sourceObjectKey":"models/v1/","starterCell":"from google.cloud import storage\nclient = storage.Client.create_anonymous_client()\nbucket = client.bucket('neuralgcm')\nblob = bucket.blob('models/v1/')\nwith blob.open(\"rb\") as fh:\n    print(fh.read(64))","starterSamplePath":"deterministic_0_7_deg.pkl","totalObjects":"flat keyspace (count unknown from bounded probe)"},"summary":"Trained weights for NeuralGCM, the hybrid model of Kochkov et al. (Nature, 2024) that couples a differentiable dynamical core for atmospheric dynamics with learned neural network…","title":"NeuralGCM Pre-trained Model Checkpoints (v1)","updatedAt":"2026-11-14T17:48:41.566Z","visibility":"public","websiteUrl":"https://neuralgcm.readthedocs.io/en/latest/checkpoints.html"}}