Gemini 4 Hydrogen (High): benchmark results

Google DeepMind model adapted from Gemini 4 Carbon, released publicly through the Gemini API and apps on November 12, 2026. Low, medium, high (default) and xhigh thinking levels; text, image, audio and video inputs. Provider: Google. Released 2026-11-12. Access: API.

Aggregate Rating 1948 ± 14, rank #20 of 2205 rated models, from 66 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AutomationBench-AA - Operations - Raw Objective Completion97.6Raw Objective Completion (%)100
AA GDP.pdf - Insurance - Criterion Pass Rate90.75Criterion Pass Rate (%)98.8
AutomationBench-AA - Finance - Raw Objective Completion94.91Raw Objective Completion (%)98.8
AutomationBench-AA74.44Guardrail-Adjusted Objective Completion (%)98.1
AutomationBench-AA - HR - Raw Objective Completion81.24Raw Objective Completion (%)97.9
AA-Omniscience v1.1 Hallucination Rate8.68Hallucination Rate (%)97.8
AA GDP.pdf - Criterion Pass Rate81.7Macro Criterion Pass Rate (%)97.7
AutomationBench-AA - Strict47.18Strict Task Success (%)97.7
AutomationBench51.14Strict Task Pass Rate (%)97.4
AA GDP.pdf - Construction - Criterion Pass Rate78.54Criterion Pass Rate (%)95.3
AA GDP.pdf - Engineering - Criterion Pass Rate82.27Criterion Pass Rate (%)95.3
AA GDP.pdf30All Criteria Pass Rate (%)95.2

Interactive version: theaggregate.ai/model?slug=gemini-4-hydrogen-high · How It Works · Data refreshed daily, snapshot 2026-11-13.