Frontier Log

Gemini 4 Argon vs Gemini 4 Hydrogen

Gemini 4 ArgonGemini 4 Hydrogen
Announced
Released
Input / 1M tokens$2$4 from 1 Dec 2026$1.50
Cached input / 1M tokens$0.10$0.20 from 1 Dec 2026$0.075
Output / 1M tokens$10$20 from 1 Dec 2026$7.50
Context window1M1M
Output limit65K65K
Benchmarks
Terminal-Bench Science 0.157.6%51.4%1
OSWorld 2.1 Offline set (82 tasks), partial score74.7%277.6%3
Chartography No tools71.6%466.6%5
Agents' Last Exam v1.1 Score, all tasks (122)69.0%671.6%7

Download image

Gemini 4 ArgonGemini 4 Hydrogen
Other benchmarks
HLE-Diamond No toolsNo data58.8%8
Terminal-Bench 4.057.4%No data
AutomationBench 1.0.6 Private set51.3%9No data
AutomationBench 1.0.6 AutomationBench-AANo data76.1%10
OSWorld 2.0 (v2026.08.08) Offline set, partial score69.2%11No data
FrontierSWE v255.0%12No data
DeepSWE v1.177.9%No data
SEC-bench Pro 26050583.4%13No data
Agents' Last Exam v1 Pass rate, set not stated39.5%14No data
GraphWalks BFS, up to 128K99.7%15No data
GraphWalks BFS, 256K + 1M problems84.2%No data

Notes

  1. Terminal-Bench Science 0.1, Gemini 4 Hydrogen: mini-swe-agent vision harness · highest thinking · average of 5 attempts per task
  2. OSWorld 2.1, Offline set (82 tasks), partial score, Gemini 4 Argon: Gemini CUA harness · highest thinking · max of 3 runs, one attempt per run
  3. OSWorld 2.1, Offline set (82 tasks), partial score, Gemini 4 Hydrogen: Gemini CUA harness · highest thinking · max of 3 runs, one attempt per run
  4. Chartography, No tools, Gemini 4 Argon: Run by Surge AI · Surge AI's public leaderboard
  5. Chartography, No tools, Gemini 4 Hydrogen: highest thinking
  6. Agents' Last Exam v1.1, Score, all tasks (122), Gemini 4 Argon: ALE-Claw harness · 5-hour window · October 17, 2026 task files · safety filters on
  7. Agents' Last Exam v1.1, Score, all tasks (122), Gemini 4 Hydrogen: Antigravity-CLI (agy) version 1.4.8 harness · highest thinking · 5-hour window · October 17, 2026 task files · safety filters on
  8. HLE-Diamond, No tools, Gemini 4 Hydrogen: Run by Artificial Analysis · highest thinking
  9. AutomationBench 1.0.6, Private set, Gemini 4 Argon: Run by Zapier · set named in the evaluation methodology the post cites
  10. AutomationBench 1.0.6, AutomationBench-AA, Gemini 4 Hydrogen: Run by Artificial Analysis · highest thinking
  11. OSWorld 2.0 (v2026.08.08), Offline set, partial score, Gemini 4 Argon: Gemini CUA harness · max of 3 runs, one attempt per run
  12. FrontierSWE v2, Gemini 4 Argon: Run by Proximal · Proximal's public leaderboard
  13. SEC-bench Pro 260505, Gemini 4 Argon: Antigravity harness · pass@1, mean of 3 attempts · 183-task JavaScript engine split (May 2026) · LLM judge for crashes in both builds
  14. Agents' Last Exam v1, Pass rate, set not stated, Gemini 4 Argon: ALE-Claw harness · 5-hour window · safety filters on
  15. GraphWalks, BFS, up to 128K, Gemini 4 Argon: 650 problems up to 128K tokens, per the model evaluation document

Related

All comparisons