Choose an institution to see its records and dossier, or choose a topic.
Through year 2027 · all records, including scheduled dates 2026 2025 2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2008 2005 1965 Show Organisations and infrastructure All types Organisation Person Document Concept Infrastructure Public body Reset filters
94 people, organisations and other records · 122 relationships · 34 claims about a valueEverything recorded up to the chosen year, whether or not still in effect · Undated records included
Claim and evidence Close ×
Gemini 1.0 training infrastructure uses hardware from ↓ Google
The Gemini v1 technical report described TPUv4 SuperPods of 4,096 chips in Gemini Ultra’s multi-datacenter training infrastructure. Copy claim link
Copy claim reference
Machine-checked, not reviewed by an editor Stated in source Supported
Supporting evidence: primary · 1 source
Date 2023-12-19 · snapshot Transaction type hardware deployment Relationship context internal Stage deployed Quantity exact · 4,096 TPUv4 chips per SuperPod (not fleet total) Scope and limits 4,096 is the per-SuperPod configuration, not the fleet total or number actively training. Standby cubes were retained. Internal hardware use does not establish an external commercial transaction.
The evidence 01 primary source Supports the claim
Gemini: A Family of Highly Capable Multimodal Models ↗ Gemini Team, Google · 2023-12-19
Copy source reference
Cite the original source Cite the original document, not Selekoa. Check these details against it.
Copy APA
Download BibTeX Download RIS Download CSL-JSON
Paraphrase TPUv4 SuperPods contain 4,096 chips; some cubes were retained for standby and maintenance.
Location in source Printed page 4, section 3 Training Infrastructure, TPUv4 SuperPods paragraph
Machine-checked · 2026-09-19
Source details and check notes Authors Gemini Team, Google Version Pinned arXiv 2312.11805v1, submitted 19 December 2023; section 3 inspected directly. The mutable DeepMind PDF is a later revision and is not the historical evidence. Canonical address Open the canonical source ↗ Accessed 2026-09-19 Language English (en) Attribution Gemini Team, Google, Gemini: A Family of Highly Capable Multimodal Models (2023-12-19). Reuse No open redistribution licence asserted. Bibliographic metadata and editorial paraphrases only; no full source redistributed. Check notes Exact cited source passage inspected by a machine agent. Company statements are not independent proof of implementation. Human source review pending. Status and meaning Editorial status. Not reviewed by an editor . An exploratory record that no editor has accepted. This status does not say whether a person has read any linked passage. Each evidence record above shows how it was checked.
Relation. Supplier hardware is reported in a compute asset. No purchase price or direct purchasing contract is implied.
Date. State reported at this date; no exact start, completion date or current continuity inferred.
Record ID gemini-training-tpu-v4-2023