# Google Gemini 4 Carbon As of November 14, 2026, Google Gemini 4 Carbon ranks #2 of 118 models overall with 88.3 ±3.2 on the LLMBenchmark Index. Gemini 4 Carbon ranks #2 of 122 in reasoning, its strongest category. Google Gemini 4 Carbon costs $2.00 per million input tokens and $10.00 per million output tokens. Gemini 4 Carbon was released on November 10, 2026. Page: https://llmbenchmark.ai/models/gemini-4-carbon Data from November 14, 2026. Every figure below names the board it was read from and the day. - Maker: Google - Released: November 10, 2026 - Price per million input tokens: $2.00 - Price per million output tokens: $10.00 - Context window: 1.05M tokens - Overall rank: 2 of 118, by the LLMBenchmark Index - LLMBenchmark Index: 88.3 ±3.2 (90% margin, from 7 of 7 inputs) - Artificial Analysis Intelligence Index: 60.0 - Vibe Coding Index: 86.6 (from 4 of 6 coding benchmarks) Benchmark results (73 of 222): - LiveBench coding: 91.0% (source: LiveBench, read November 14, 2026) - LMArena WebDev: 1768 Elo (High) (source: LMArena WebDev, read November 14, 2026) - Vals Index: 70.5% (source: Vals Index, read November 14, 2026) - AA Intelligence Index: 60.0 (source: Artificial Analysis (via OpenRouter), read November 14, 2026) - Design Arena: websites: 1319 Elo (source: Design Arena, read November 14, 2026) - Design Arena: UI components: 1357 Elo (source: Design Arena, read November 14, 2026) - Design Arena: game dev: 1326 Elo (source: Design Arena, read November 14, 2026) - Design Arena: 3D: 1434 Elo (source: Design Arena, read November 14, 2026) - Design Arena: all code categories: 1348 Elo (source: Design Arena, read November 14, 2026) - Design Arena: agents building mobile apps: 1284 Elo (source: Design Arena, read November 14, 2026) - Design Arena: spreadsheets: 1224 Elo (source: Design Arena, read November 14, 2026) - LiveBench reasoning: 90.6% (source: LiveBench, read November 14, 2026) - LiveBench data analysis: 86.5% (source: LiveBench, read November 14, 2026) - LiveBench research: 92.5% (source: LiveBench, read November 14, 2026) - LiveBench browser: 80.0% (source: LiveBench, read November 14, 2026) - LiveBench long agentic: 72.7% (source: LiveBench, read November 14, 2026) - LiveBench design: 80.2% (source: LiveBench, read November 14, 2026) - Terminal-Bench 4 (Vals AI): 63.1% (Mini-SWE-agent) (source: Vals AI boards, read November 14, 2026) - Vibe Code Bench (Vals AI): 89.1% (OpenHands) (source: Vals AI boards, read November 14, 2026) - Tax Agent Bench (Vals AI): 51.8% (source: Vals AI boards, read November 14, 2026) - Legal Research Bench (Vals AI): 55.8% (source: Vals AI boards, read November 14, 2026) - Harvey's Legal Agent Benchmark (Vals AI): 17.5% (source: Vals AI boards, read November 14, 2026) - Long context reasoning (AA-LCR): 79.3% (High) (source: Artificial Analysis, read November 14, 2026) - Terminal-Bench 4.0 (Artificial Analysis run): 65.2% (High) (source: Artificial Analysis, read November 14, 2026) - FrontierCode 2.0: 40.6% (High) (source: Artificial Analysis, read November 14, 2026) - HLE-Diamond (Artificial Analysis run): 69.1% (High) (source: Artificial Analysis, read November 14, 2026) - GDPval-AA: 1733 Elo (High) (source: Artificial Analysis, read November 14, 2026) - AA-Omniscience: 53.5 (High) (source: Artificial Analysis, read November 14, 2026) - CritPt v2, hard subset (Artificial Analysis run): 69.0% (High) (source: Artificial Analysis, read November 14, 2026) - FrontierSWE: 61.2% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - APEX-Agents: 84.2% (source: Epoch AI benchmarking hub, read November 14, 2026) - GPQA Diamond (Epoch AI run): 96.5% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - ARC-AGI: 98.5% (Medium) (source: Epoch AI benchmarking hub, read November 14, 2026) - ARC-AGI-2: 94.2% (Medium) (source: Epoch AI benchmarking hub, read November 14, 2026) - SimpleBench: 92.6% (source: Epoch AI benchmarking hub, read November 14, 2026) - Chess puzzles: 78.0% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - Mystery game puzzles: 73.0% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - EBR-bench: 87.6% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - OTIS Mock AIME: 100.0% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - FrontierMath, tiers 1 to 3: 90.5% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - FrontierMath, tier 4: 95.1% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - SimpleQA Verified: 77.7% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - Furniture assembly: 77.5% (High) (source: Epoch AI benchmarking hub, read November 14, 2026) - ALE-Bench: 2374.7 (High) (source: ALE-Bench (Sakana AI), read November 14, 2026) - Vending-Bench 2: $16,484 (source: Epoch AI benchmarking hub, read November 14, 2026) - Epoch Capabilities Index: 169.8 (source: Epoch AI benchmarking hub, read November 14, 2026) - LMArena Text: 1538 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: coding: 1564 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: hard prompts: 1551 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: instruction following: 1539 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: creative writing: 1533 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: English: 1530 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: non-English: 1534 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: Chinese: 1613 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: Russian: 1560 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: multi-turn: 1569 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: longer queries: 1549 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: expert: 1582 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: hard prompts in English: 1545 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: software and IT: 1552 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: mathematical work: 1538 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: science: 1535 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: medicine and healthcare: 1525 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: legal and government: 1542 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: business and finance: 1520 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: writing and language: 1542 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena Text: entertainment and media: 1527 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena WebDev: React: 1792 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - LMArena WebDev: HTML: 1770 Elo (High) (source: LMArena (Hugging Face dataset), read November 14, 2026) - FACTS Benchmark Suite: 78.4% (source: Kaggle Benchmarks, read November 14, 2026) - MathArena: 85.4% (high) (source: MathArena, read November 14, 2026) - PRBench Legal: 53.3% (source: Scale SEAL leaderboards, read November 14, 2026) - CursorBench 4.0: 60.4% (high) (source: CursorBench (Cursor), read November 14, 2026)