- AA Index v5.0 59
- Vals Index v2.2 70.4
- SWE-sweep 9.6
- FrontierSWE 2 61.2
- Terminal-Bench 4.0 (AA) 65.2
- Agents' Last Exam v1.1 (Score) 77.4
- OSWorld-2.1 (Offline subset) 81.5
- AutomationBench (Zapier) 53.9
- Vending-Bench 2 (Bank balance $) 16484
- Chartography 72.5
- Blueprint-Bench 2 (Similarity score) 58.6
- GraphWalks (256k to 1M, BFS (F1)) 96.2
- RiemannBench 80
- Terminal-Bench-Science 0.1 (AA) 64.8
- CWE-bench 71
- ExploitGym 54.8
- SEC-bench Pro 88.2
- WebDev Arena (#4, 1768, text) 1768
- Gray Swan Indirect Prompt Injection (IPI) Benchmark, attack success rate 0.0% (lower is better) 0
Gemini 4 Carbon
Gemini 4 Carbon ranks first on 14 of the 29 benchmarks it has scores on.
Google DeepMind Released 28 Oct 2026 35
Benchmarks with a score
14
Ranked first, out of five or more models
19
In the top three
Google DeepMind Oct 2026 Gemini 4 Carbon
Closed weights
No Epoch Capabilities Index score yet
- AutomationBench 53.9% #1 /5
- Agents' Last Exam 45.9% #1 /12
- CritPt 36% #1 /160
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LMArena (Elo) | 1,528 | #1 of 333 | Official leaderboard | |
| GraphWalks | 96.2% | #4 of 21 | Model card |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CritPt | 36% | #1 of 160 | Epoch AI | |
| Terminal-Bench-Science | 64.8% | #1 of 9 | Model card | |
| Riemann-bench | 80% | #1 of 4 | Model card |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Code Arena | 1,768 | #4 of 124 | Official leaderboard | |
| Vibe Code Bench | 89.1% | #8 of 115 | Official leaderboard | |
| FrontierCode 2.0 | 40% | #4 of 26 | Official leaderboard | |
| FrontierCS 2 | 54.2% | #3 of 16 | Official leaderboard | |
| FrontierSWE | 61.2% | #4 of 9 | Model card | |
| WebDev Arena | 1,768 | #1 of 2 | Model card | |
| SWE-sweep | 9.60% | #1 of 1 | Model card |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Agent Arena | 0.16 | #2 of 53 | Official leaderboard | |
| Terminal-Bench 4.0 | 65.2% | #4 of 23 | Model card | |
| AutomationBench | 53.9% | #1 of 5 | Official leaderboard | |
| OSWorld 2.1 (offline subset) | 81.5% | #1 of 2 | Model card |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Legal Research Bench | 55.8% | #1 of 88 | Official leaderboard | |
| Tax Agent Bench | 51.8% | #1 of 79 | Official leaderboard | |
| Vending-Bench 2 | $16,484 | #1 of 66 | Official leaderboard | |
| Vals Finance Index | 67% | #1 of 35 | Official leaderboard | |
| Vending-Bench Arena | $15,962 | #1 of 32 | Official leaderboard | |
| Excel Modeling Benchmark v1.1 | 76.5% | #2 of 37 | Official leaderboard | |
| Vals Index v2.2 | 70.5% | #2 of 32 | Official leaderboard | |
| Agents' Last Exam | 45.9% | #1 of 12 | Official leaderboard | |
| Vals Legal Index | 75.9% | #3 of 33 | Official leaderboard | |
| Vals Finance Agent v2.1 | 67.2% | #5 of 35 | Official leaderboard | |
| Harvey LAB v1.1 (HLAB-Vals-1.1) | 17.5% | #7 of 33 | Official leaderboard | |
| Agents' Last Exam v1.1 | 77.4% | #1 of 1 | Model card |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Blueprint-Bench 2 | 58.6% | #1 of 35 | Official leaderboard | |
| Chartography | 72.5% | #6 of 12 | Model card |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| ExploitGym | 54.8% | #1 of 14 | Model card | |
| CWE-bench | 71% | #1 of 12 | Official leaderboard | |
| SEC-bench Pro | 88.2% | #1 of 10 | Model card | |
| Gray Swan IPI Arena benchmark | 0% | #10 of 10 | Model card |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Artificial Analysis Intelligence Index v5 | 59.0 | #1 of 1 | Model card |
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.