{"paper":{"title":"Chronos-2: From Univariate to Universal Forecasting","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Chronos-2 is a pretrained model that performs zero-shot forecasting on univariate, multivariate, and covariate-informed tasks via group attention for in-context learning.","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Abdul Fatir Ansari, Andreas Auer, Boran Han, Danielle C. Maddix, George Karypis, Hao Wang, Huibin Shen, Huzefa Rangwala, Jaris K\\\"uken, Junming Yin, Lorenzo Stella, Michael Bohlke-Schneider, Mononito Goswami, Nick Erickson, Oleksandr Shchur, Pablo Guerron, Pedro Mercado, Prateek Mutalik Desai, Shubham Kapoor, Syama Sundar Rangapuram, Tony Hu, Xiyuan Zhang, Yuyang Wang","submitted_at":"2025-10-17T17:00:53Z","abstract_excerpt":"Pretrained time series models have enabled inference-only forecasting systems that produce accurate predictions without task-specific training. However, existing approaches largely focus on univariate forecasting, limiting their applicability in real-world scenarios where multivariate data and covariates play a crucial role. We present Chronos-2, a pretrained model capable of handling univariate, multivariate, and covariate-informed forecasting tasks in a zero-shot manner. Chronos-2 employs a group attention mechanism that facilitates in-context learning (ICL) through efficient information sha"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Chronos-2 delivers state-of-the-art performance across three comprehensive benchmarks: fev-bench, GIFT-Eval, and Chronos Benchmark II. On fev-bench, which emphasizes multivariate and covariate-informed forecasting, Chronos-2's universal ICL capabilities lead to substantial improvements over existing models. On tasks involving covariates, it consistently outperforms baselines by a wide margin.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That training exclusively on synthetic datasets that impose diverse multivariate structures on univariate series will produce a model whose in-context learning generalizes to real-world multivariate and covariate distributions without domain-specific fine-tuning.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Chronos-2 adds group attention to a pretrained time series model so it can do zero-shot forecasting on univariate, multivariate, and covariate tasks by learning from synthetic data that imposes multivariate structure on univariate series.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Chronos-2 is a pretrained model that performs zero-shot forecasting on univariate, multivariate, and covariate-informed tasks via group attention for in-context learning.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"cfeb3e3e3f87fbc782e3bd6e00f12fb47fae3f0e95f8bdb5e36acf269eb258b1"},"source":{"id":"2510.15821","kind":"arxiv","version":1},"verdict":{"id":"1170fea0-5829-416e-8693-adc371965820","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T01:16:12.885791Z","strongest_claim":"Chronos-2 delivers state-of-the-art performance across three comprehensive benchmarks: fev-bench, GIFT-Eval, and Chronos Benchmark II. On fev-bench, which emphasizes multivariate and covariate-informed forecasting, Chronos-2's universal ICL capabilities lead to substantial improvements over existing models. On tasks involving covariates, it consistently outperforms baselines by a wide margin.","one_line_summary":"Chronos-2 adds group attention to a pretrained time series model so it can do zero-shot forecasting on univariate, multivariate, and covariate tasks by learning from synthetic data that imposes multivariate structure on univariate series.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That training exclusively on synthetic datasets that impose diverse multivariate structures on univariate series will produce a model whose in-context learning generalizes to real-world multivariate and covariate distributions without domain-specific fine-tuning.","pith_extraction_headline":"Chronos-2 is a pretrained model that performs zero-shot forecasting on univariate, multivariate, and covariate-informed tasks via group attention for in-context learning."},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"fdac3f6c2883ac05eb3b9f96a5ad70b7807745caa87a0287e60831ee245faa2a"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}