Skip to content
Agents tracked: 492 Releases (7d): 545 Agent status: 1 with issuesSince last limit reset: Claude 9d · Codex 7d For AI agents: Markdown Updated Nov 20, 2026

Gemini 4 Neon: benchmarks as reported by Google DeepMind

The launch table Google DeepMind published for Gemini 4 Neon on Nov 18, 2026, re-checked against the page daily. Independent results for Gemini 4 Neon on the benchmarks hub as they appear.

Gemini 4 Neon: benchmarks as reported by Google DeepMind

From Gemini 4 Neon model card, published Nov 18, 2026. Each number was re-read on that page on Nov 19, 2026.

BenchmarkGemini 4 NeonClaude Opus 5.5GPT-6.2 Sol
Capability indices
AA Index v5.0.1 61 5957
Vals Index v2.2 70.9% 70.1%66.1%
Agentic coding
CursorBench 4.1Score 65.4% 58.1%—
Terminal-bench 4.1Cheating adjusted accuracy 65.2% 65.8%61.5%
FrontierCode 2.0Main score 46.2% 47.3%43.8%
SWE-sweepProvider harness 39.0% 10.1%8.3%
SWE-sweepMini-SWE harness 12.8% 3.6%11.6%
Knowledge work
AutomationBench 56.3% 42.5%41.3%
Agents' Last Exam v1.1 75.8% 71.6%73.0%
Science and math
RiemannBench 84.0% 80.0%80.0%
Terminal-Bench-Science 0.1 70.5% 63.3%68.0%
Reasoning
HLE-Diamond 71.3% 67.9%64.0%
Multimodal understanding
MKAClimb WSR 56.5% 53.0%52.7%
Chartography 73.5% 66.3%75.1%
LVBench 89.0% 83.7%87.6%
Long context
GraphWalks256k to 1M, BFS (F1) 95.2% 66.8%80.1%
Cybersecurity
CWE-Bench v2 70.7% 67.1%66.0%
SEC-bench Pro 90.5% —88.3%
WebExploitBench v1.2 51.3% 37.1%—
As reported by Google DeepMind: Scores as published by Google DeepMind in the model card; methodology at deepmind.google/models/evals-methodology/gemini-4-neon. None of these tests is in Gid Bench. These numbers are not part of Gid Bench, which uses independent runs only.

More