# OpenAI GPT-6.1 Terra

As of October 29, 2026, OpenAI GPT-6.1 Terra ranks #9 of 111 models overall with 76.8 ±5.8 on the LLMBenchmark Index. GPT-6.1 Terra ranks #1 of 106 in math, its strongest category. OpenAI GPT-6.1 Terra costs $1.00 per million input tokens and $5.00 per million output tokens. GPT-6.1 Terra was released on October 26, 2026.

Page: https://llmbenchmark.ai/models/gpt-6-1-terra
Data from October 29, 2026. Every figure below names the board it was read from and the day.

- Maker: OpenAI
- Released: October 26, 2026
- Price per million input tokens: $1.00
- Price per million output tokens: $5.00
- Context window: 1.05M tokens
- Overall rank: 9 of 111, by the LLMBenchmark Index
- LLMBenchmark Index: 76.8 ±5.8 (90% margin, from 6 of 7 inputs)
- Artificial Analysis Intelligence Index: 51.0
- Vibe Coding Index: 77.4 (from 4 of 6 coding benchmarks)

Benchmark results (36 of 206):

- LiveBench coding: 94.3% (source: LiveBench, read October 29, 2026)
- LMArena WebDev: 1673 Elo (Max) (source: LMArena WebDev, read October 29, 2026)
- Vals Index: 60.5% (source: Vals Index, read October 29, 2026)
- AA Intelligence Index: 51.0 (source: Artificial Analysis (via OpenRouter), read October 29, 2026)
- Design Arena: websites: 1283 Elo (source: Design Arena, read October 29, 2026)
- Design Arena: UI components: 1277 Elo (source: Design Arena, read October 29, 2026)
- Design Arena: game dev: 1348 Elo (source: Design Arena, read October 29, 2026)
- Design Arena: all code categories: 1295 Elo (source: Design Arena, read October 29, 2026)
- LiveBench reasoning: 88.7% (source: LiveBench, read October 29, 2026)
- LiveBench data analysis: 82.3% (source: LiveBench, read October 29, 2026)
- LiveBench research: 73.9% (source: LiveBench, read October 29, 2026)
- LiveBench browser: 93.1% (source: LiveBench, read October 29, 2026)
- LiveBench long agentic: 48.6% (source: LiveBench, read October 29, 2026)
- LiveBench design: 82.3% (source: LiveBench, read October 29, 2026)
- Terminal-Bench 4 (Vals AI): 49.5% (Mini-SWE-agent) (source: Vals AI boards, read October 29, 2026)
- Vibe Code Bench (Vals AI): 84.0% (OpenHands) (source: Vals AI boards, read October 29, 2026)
- Tax Agent Bench (Vals AI): 15.2% (source: Vals AI boards, read October 29, 2026)
- Legal Research Bench (Vals AI): 26.0% (source: Vals AI boards, read October 29, 2026)
- Harvey's Legal Agent Benchmark (Vals AI): 1.7% (source: Vals AI boards, read October 29, 2026)
- Humanity's Last Exam (Artificial Analysis run): 46.2% (Max) (source: Artificial Analysis, read October 29, 2026)
- SciCode: 55.1% (Max) (source: Artificial Analysis, read October 29, 2026)
- Long context reasoning (AA-LCR): 82.7% (Max) (source: Artificial Analysis, read October 29, 2026)
- APEX-Agents: 68.3% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- GPQA Diamond (Epoch AI run): 94.3% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- Mystery game puzzles: 74.0% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- EBR-bench: 78.1% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- OTIS Mock AIME: 100.0% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- FrontierMath, tiers 1 to 3: 90.9% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- FrontierMath, tier 4: 92.7% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- SimpleQA Verified: 57.3% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- Furniture assembly: 79.2% (Max) (source: Epoch AI benchmarking hub, read October 29, 2026)
- ALE-Bench: 2608.8 (Max) (source: ALE-Bench (Sakana AI), read October 29, 2026)
- LMArena WebDev: React: 1657 Elo (Max) (source: LMArena (Hugging Face dataset), read October 29, 2026)
- LMArena WebDev: HTML: 1719 Elo (Max) (source: LMArena (Hugging Face dataset), read October 29, 2026)
- MultiChallenge: 74.2% (source: Scale SEAL leaderboards, read October 29, 2026)
- Humanity's Last Exam (Scale run): 35.7% (source: Scale SEAL leaderboards, read October 29, 2026)