Gemini 4 Hydrogen benchmarks
Every row below is a published figure, not a test run by this site. Google’s model card (results as of November 2026) is one table; the 12 November 2026 announcement blog adds Artificial Analysis 51 / 54, Agent Arena #5, and named Vals scores; Vals AI and OpenRouter publish their own numbers. Do not mix benchmark editions.
Source: Google DeepMind, 12 November 2026 · Source: Google Blog, 12 November 2026 · Source: Vals AI (independent), Retrieved 13 November 2026 · Source: OpenRouter (independent), Retrieved 13 November 2026
| Benchmark | Edition / notes | Hydrogen score | Setting | Who measured | Date | Link |
|---|---|---|---|---|---|---|
| SWE-sweep | as published | 13.5% | — | Google DeepMind model card | Results as of November 2026 | Source |
| Terminal-Bench 4.1 | cheating-adjusted | 57.0% | — | Google DeepMind model card | Results as of November 2026 | Source |
| Terminal-Bench 4.1 | vs Gemini 3.8 Flash (18.2%) | 57% | high default (blog wording) | Google announcement blog | 12 November 2026 | Source |
| Agents’ Last Exam v1.1 | score | 71.6% | — | Google DeepMind model card | Results as of November 2026 | Source |
| AutomationBench-AA | as published | 76.1% | — | Google DeepMind model card | Results as of November 2026 | Source |
| Vending-Bench 2 | bank balance ($) | $14,549 | — | Google DeepMind model card | Results as of November 2026 | Source |
| OSWorld-2.1 | offline subset | 77.6% | — | Google DeepMind model card | Results as of November 2026 | Source |
| MKA | Climb WSR accuracy | 53.4% | — | Google DeepMind model card | Results as of November 2026 | Source |
| Chartography | as published | 66.6% | — | Google DeepMind model card | Results as of November 2026 | Source |
| Terminal-Bench-Science 0.1 | as published | 51.4% | — | Google DeepMind model card | Results as of November 2026 | Source |
| HLE-Diamond | as published | 58.8% | — | Google DeepMind model card; also named on the announcement blog | Results as of November 2026 / 12 November 2026 | Source |
| Artificial Analysis Intelligence Index | as published by Google | 51 | default high thinking | Google announcement blog | 12 November 2026 | Source |
| Artificial Analysis Intelligence Index | as published by Google | 54 | xhigh thinking | Google announcement blog | 12 November 2026 | Source |
| Agent Arena | live benchmark, arena.ai (Google’s report) | #5 | — | Google announcement blog | 12 November 2026 | Source |
| Vals Tax Agent Bench | Google’s report of a Vals figure | 52.3% (#1 overall) | — | Google announcement blog | 12 November 2026 | Source |
| Vals Legal Research Bench | Google’s report of a Vals figure | 55.8% (tied #1) | — | Google announcement blog | 12 November 2026 | Source |
| Vals Index | Vals Index (Vals’s page lists v2.2 in its write-up) | 68.49% ± 0.99 (#6 of 37) | default (Vals page) | Vals AI (independent) | Retrieved 13 November 2026 | Source |
| Tax Agent Bench | Vals’s own page | 52.30% ± 3.60 (#1 of 85) | default (Vals page) | Vals AI (independent) | Retrieved 13 November 2026 | Source |
| Terminal-Bench 4.0 | Vals’s own page; not the same edition as Google’s 4.1 | 54.04% ± 3.31 (#11 of 64) | default (Vals page) | Vals AI (independent) | Retrieved 13 November 2026 | Source |
| Vibe Code Bench v1.1 | Vals’s own page | 87.78% ± 2.29 (#13 of 128) | default (Vals page) | Vals AI (independent) | Retrieved 13 November 2026 | Source |
| Models Arena Game Development Elo | OpenRouter summary table, source column “Design Arena” | 1357 | — | OpenRouter (independent) | Retrieved 13 November 2026 | Source |
Google’s model-card table also reports Claude Sonnet 5.5 and GPT-6.1 Terra in the same cells; those comparison columns are on the compare page. We do not invent a Hydrogen-versus-Sol official table. A separate Anthropic Terminal-Bench 4.0 number for Sonnet 5.5 is noted there and is not copied into this Hydrogen score list.