Frontier Log

Gemini 4 Carbon vs Gemini 4 Hydrogen

Gemini 4 CarbonGemini 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 0.164.8%151.4%2
OSWorld 2.1 Offline set (82 tasks), partial score81.5%377.6%4
Chartography No tools72.5%566.6%6
Agents' Last Exam v1.1 Score, all tasks (122)77.4%771.6%8

Download image

Gemini 4 CarbonGemini 4 Hydrogen
Other benchmarks
HLE-Diamond No toolsNo data58.8%9
Terminal-Bench 4.065.2%10No data
AutomationBench 1.0.6 Private set53.9%11No data
AutomationBench 1.0.6 AutomationBench-AANo data76.1%12
FrontierSWE v261.2%13No data
SEC-bench Pro 26050588.2%14No data
GraphWalks BFS, 256K + 1M problems96.2%No data

Notes

  1. Terminal-Bench Science 0.1, Gemini 4 Carbon: Run by Artificial Analysis · mini-swe-agent harness · high thinking · average of 3 runs
  2. Terminal-Bench Science 0.1, Gemini 4 Hydrogen: mini-swe-agent vision harness · highest thinking · average of 5 attempts per task
  3. OSWorld 2.1, Offline set (82 tasks), partial score, Gemini 4 Carbon: Gemini CUA harness · highest thinking · max of 3 runs, one attempt per run
  4. 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
  5. Chartography, No tools, Gemini 4 Carbon: Run by Surge AI · Surge AI's public leaderboard
  6. Chartography, No tools, Gemini 4 Hydrogen: highest thinking
  7. Agents' Last Exam v1.1, Score, all tasks (122), Gemini 4 Carbon: ALE-Claw harness · 5-hour window · October 17, 2026 task files · safety filters on
  8. 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
  9. HLE-Diamond, No tools, Gemini 4 Hydrogen: Run by Artificial Analysis · highest thinking
  10. Terminal-Bench 4.0, Gemini 4 Carbon: Run by Artificial Analysis · mini-swe-agent harness · high thinking · average of 3 runs
  11. AutomationBench 1.0.6, Private set, Gemini 4 Carbon: Run by Zapier · set named in the evaluation methodology the post cites
  12. AutomationBench 1.0.6, AutomationBench-AA, Gemini 4 Hydrogen: Run by Artificial Analysis · highest thinking
  13. FrontierSWE v2, Gemini 4 Carbon: Run by Proximal · Proximal's public leaderboard
  14. SEC-bench Pro 260505, Gemini 4 Carbon: Antigravity harness · pass@1, mean of 3 attempts · 183-task JavaScript engine split (May 2026) · LLM judge for crashes in both builds

Related

All comparisons