EarthRecord
Agents analyze multidecade satellite observations to answer land-change questions. Reference checks score answers to 238 questions across 56 sites.
The benchmark requires tool use and multisensor temporal reasoning. Perturbation controls test sensitivity to the evidence rather than only geographic familiarity.
- Made by
- AIEyes team
- Released
- November 2026
- Kind
- Lab and data work
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Subject
- Earth and environmental sciences
- Language
- English
- In and out
- Text, Maps and geodata, Time series, Images → Text, Actions
- Interaction
- Agentic
- Environment
- APIs and tools
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
- GPT-6 Astra #4 on the capability index
- Last model added
- October 2026
- Scores refreshed
- November 18th, 2026
In the authors’ words
While multimodal agents now carry out complex data analysis with tools, their ability to turn decades of satellite imagery into reliable land-change analysis, crucial for Earth observation science, remains unmeasured. Existing benchmarks fall short: they pose questions on curated images or give agents preprocessed inputs, leaving out the raw multi-sensor archives where such analysis begins. We introduce EarthRecord, a benchmark built on the raw record of 56 sites: four decades of Landsat, MODIS, VIIRS, and Sentinel-2 data, with cloud masking and cross-sensor harmonization left to the agent. It poses 238 land-change questions across ten analysis tasks, scored against withheld references from global products, agreement among methods, and visual interpretation. Two controls test what answers rest on: the closed book removes the record, and instrument pairs alter one satellite's data while nothing changes on the ground. Evaluations of twelve multimodal agents reveal that leading agents complete most analyses end to end but falter when an answer requires mapping from imagery. Further analysis shows that the record moves answers from prior knowledge to the pixels but not yet to the real change on the ground: when one satellite's data are altered, half of the right answers follow the alteration. EarthRecord thus provides a principled platform for measuring and advancing agents from pixels to land change.
Claude Opus 5.5 leads with 85.4% Claude Opus 5.5 leads with 85.4%
Best score per entry from the benchmark's own leaderboard, read on November 18th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
- 1 Claude Opus 5.5 85.4%
- 2 GPT-6 Sol 84.1%
- 3 GPT-6.1 Sol 81.4%
- 4 Claude Sonnet 5.5 81%
- 5 GPT-6 Astra 77%
- 6 GPT-6.1 Terra 73.5%
- 7
Gemini 3 Flash 72.6%
- 8 DeepSeek Flash 68.1%
- 9 Kimi K2.5 54%
- 10 Claude Haiku 5.5 41.2%
- 11
Qwen3.7 Flash 19.5%
- 12 GPT-6 Luna 18.6%
Source: Official EarthRecord leaderboard
- 1 Claude Opus 5.5 85.4%
- 2 GPT-6 Sol 84.1%
- 3 GPT-6.1 Sol 81.4%
- 4 Claude Sonnet 5.5 81%
- 5 GPT-6 Astra 77%
- 6 GPT-6.1 Terra 73.5%
- 7
Gemini 3 Flash 72.6%
- 8 Kimi K2.5 54%
- 9 Claude Haiku 5.5 41.2%
- 10
Qwen3.7 Flash 19.5%
- 11 GPT-6 Luna 18.6%
Source: Official EarthRecord leaderboard
| # | Entry | Score ± | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | 85.4% 80.5 to 89.8 | Sep 2026 | Official leaderboard | |
| 2 | GPT-6 Sol | 84.1% ±4.9 | Sep 2026 | Official leaderboard | |
| 3 | GPT-6.1 Sol | 81.4% 76.1 to 86.3 | Sep 2026 | Official leaderboard | |
| 4 | Claude Sonnet 5.5 | 81% ±5.3 | Sep 2026 | Official leaderboard | |
| 5 | GPT-6 Astra | 77% 71.2 to 82.3 | Sep 2026 | Official leaderboard | |
| 6 | GPT-6.1 Terra | 73.5% ±5.8 | Oct 2026 | Official leaderboard | |
| 7 | | 72.6% 66.8 to 77.9 | Dec 2025 | Official leaderboard | |
| 8 | DeepSeek Flash | 68.1% ±6.2 | Official leaderboard | ||
| 9 | Kimi K2.5 | 54% 47.8 to 60.6 | Jan 2026 | Official leaderboard | |
| 10 | Claude Haiku 5.5 | 41.2% 35 to 47.8 | Oct 2026 | Official leaderboard | |
| 11 | | 19.5% 14.6 to 24.8 | Jul 2026 | Official leaderboard | |
| 12 | GPT-6 Luna | 18.6% 13.7 to 23.9 | Sep 2026 | Official leaderboard |
| # | Model | Score ± | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | 85.4% 80.5 to 89.8 | Sep 2026 | Official leaderboard | |
| 2 | GPT-6 Sol | 84.1% ±4.9 | Sep 2026 | Official leaderboard | |
| 3 | GPT-6.1 Sol | 81.4% 76.1 to 86.3 | Sep 2026 | Official leaderboard | |
| 4 | Claude Sonnet 5.5 | 81% ±5.3 | Sep 2026 | Official leaderboard | |
| 5 | GPT-6 Astra | 77% 71.2 to 82.3 | Sep 2026 | Official leaderboard | |
| 6 | GPT-6.1 Terra | 73.5% ±5.8 | Oct 2026 | Official leaderboard | |
| 7 | | 72.6% 66.8 to 77.9 | Dec 2025 | Official leaderboard | |
| 8 | Kimi K2.5 | 54% 47.8 to 60.6 | Jan 2026 | Official leaderboard | |
| 9 | Claude Haiku 5.5 | 41.2% 35 to 47.8 | Oct 2026 | Official leaderboard | |
| 10 | | 19.5% 14.6 to 24.8 | Jul 2026 | Official leaderboard | |
| 11 | GPT-6 Luna | 18.6% 13.7 to 23.9 | Sep 2026 | Official leaderboard |
Rankings by part
We read 12 parts from the benchmark’s own leaderboard in 2 groups: task type and category. Each ranks models on one part of the test. Claude Opus 5.5 leads 6 of them, and 4 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Quantify
11 models we track have a score here.
- 1 GPT-6 Sol ↑1 75.5%
- 2 Claude Opus 5.5 ↓1 69.4%
- 3 Claude Sonnet 5.5 ↑1 67.3%
- 4 GPT-6.1 Sol ↓1 63.3%
- 5 GPT-6 Astra 55.1%
- 6
Gemini 3 Flash ↑1 51%
- 7 GPT-6.1 Terra ↓1 40.8%
- 8 Kimi K2.5 38.8%
- 9 Claude Haiku 5.5 24.5%
- 10 GPT-6 Luna ↑1 14.3%
Date
11 models we track have a score here.
- 1 GPT-6 Sol ↑1 65.4%
- 2 GPT-6.1 Sol ↑1 65.4%
- 3 Claude Sonnet 5.5 ↑1 61.5%
- 4 GPT-6.1 Terra ↑2 57.7%
- 5
Gemini 3 Flash ↑2 53.8%
- 6 GPT-6 Astra ↓1 53.8%
- 7 Claude Opus 5.5 ↓6 53.8%
- 8 Kimi K2.5 46.2%
- 9 GPT-6 Luna ↑2 19.2%
- 10 Claude Haiku 5.5 ↓1 15.4%
Anomaly and climate
11 models we track have a score here.
- 1 Claude Opus 5.5 84.2%
- 2 GPT-6 Sol 78.9%
- 3 Claude Sonnet 5.5 ↑1 78.9%
- 4 GPT-6.1 Sol ↓1 73.7%
- 5 GPT-6.1 Terra ↑1 73.7%
- 6 GPT-6 Astra ↓1 68.4%
- 7
Gemini 3 Flash 63.2%
- 8 Claude Haiku 5.5 ↑1 42.1%
- 9 Kimi K2.5 ↓1 36.8%
- 10 GPT-6 Luna ↑1 15.8%
Detect
11 models we track have a score here.
- 1 Claude Opus 5.5 100%
- 2 GPT-6 Sol 87.5%
- 3 Claude Sonnet 5.5 ↑1 87.5%
- 4 GPT-6 Astra ↑1 81.2%
- 5 GPT-6.1 Sol ↓2 81.2%
- 6 GPT-6.1 Terra 75%
- 7
Gemini 3 Flash 62.5%
- 8 Kimi K2.5 56.2%
- 9
Qwen3.7 Flash ↑1 31.2%
- 10 Claude Haiku 5.5 ↓1 25%
Trend
11 models we track have a score here.
- 1 Claude Opus 5.5 96.4%
- 2 GPT-6 Sol 89.3%
- 3 Claude Sonnet 5.5 ↑1 89.3%
- 4 GPT-6.1 Sol ↓1 89.3%
- 5 GPT-6.1 Terra ↑1 85.7%
- 6 GPT-6 Astra ↓1 78.6%
- 7
Gemini 3 Flash 75%
- 8 Kimi K2.5 75%
- 9 Claude Haiku 5.5 60.7%
- 10
Qwen3.7 Flash 39.3%
Recovery
11 models we track have a score here.
- 1 GPT-6 Astra ↑4 100%
- 2 GPT-6 Sol 100%
- 3 GPT-6.1 Sol 100%
- 4 GPT-6.1 Terra ↑2 90.9%
- 5
Gemini 3 Flash ↑2 81.8%
- 6 Claude Opus 5.5 ↓5 81.8%
- 7 Claude Sonnet 5.5 ↓3 81.8%
- 8 Kimi K2.5 72.7%
- 9
Qwen3.7 Flash ↑1 54.5%
- 10 Claude Haiku 5.5 ↓1 36.4%
Season
11 models we track have a score here.
- 1
Gemini 3 Flash ↑6 100%
- 2 Claude Opus 5.5 ↓1 100%
- 3 GPT-6 Astra ↑2 91.7%
- 4 GPT-6 Sol ↓2 91.7%
- 5 GPT-6.1 Terra ↑1 91.7%
- 6 Claude Sonnet 5.5 ↓2 83.3%
- 7 GPT-6.1 Sol ↓4 83.3%
- 8 Claude Haiku 5.5 ↑1 83.3%
- 9 Kimi K2.5 ↓1 58.3%
- 10 GPT-6 Luna ↑1 41.7%
Compare
11 models we track have a score here.
- 1 Claude Opus 5.5 100%
- 2 GPT-6 Astra ↑3 95.8%
- 3 GPT-6.1 Sol 91.7%
- 4
Gemini 3 Flash ↑3 87.5%
- 5 Claude Sonnet 5.5 ↓1 87.5%
- 6 GPT-6.1 Terra 83.3%
- 7 GPT-6 Sol ↓5 79.2%
- 8 Kimi K2.5 58.3%
- 9 Claude Haiku 5.5 50%
- 10
Qwen3.7 Flash 37.5%
Attribute
11 models we track have a score here.
- 1
Gemini 3 Flash ↑6 100%
- 2 GPT-6 Astra ↑3 100%
- 3 Claude Opus 5.5 ↓2 100%
- 4 GPT-6 Sol ↓2 100%
- 5 GPT-6.1 Sol ↓2 100%
- 6 Claude Sonnet 5.5 ↓2 95%
- 7 GPT-6.1 Terra ↓1 95%
- 8 Kimi K2.5 60%
- 9 Claude Haiku 5.5 50%
- 10 GPT-6 Luna ↑1 45%
Regional statistics
11 models we track have a score here.
- 1 Claude Opus 5.5 100%
- 2 GPT-6 Sol 100%
- 3 Claude Sonnet 5.5 ↑1 100%
- 4 GPT-6.1 Sol ↓1 100%
- 5 GPT-6.1 Terra ↑1 100%
- 6
Gemini 3 Flash ↑1 95.2%
- 7 GPT-6 Astra ↓2 95.2%
- 8 Kimi K2.5 61.9%
- 9 Claude Haiku 5.5 57.1%
- 10
Qwen3.7 Flash 0%
Computable
11 models we track have a score here.
- 1 GPT-6.1 Sol ↑2 89.5%
- 2 GPT-6 Astra ↑3 87.6%
- 3 Claude Opus 5.5 ↓2 87.6%
- 4 GPT-6 Sol ↓2 86.9%
- 5 GPT-6.1 Terra ↑1 86.3%
- 6 Claude Sonnet 5.5 ↓2 85%
- 7
Gemini 3 Flash 76.5%
- 8 Kimi K2.5 62.7%
- 9 Claude Haiku 5.5 49.7%
- 10
Qwen3.7 Flash 20.9%
Recognition
11 models we track have a score here.
- 1 Claude Opus 5.5 80.8%
- 2 GPT-6 Sol 78.1%
- 3 Claude Sonnet 5.5 ↑1 72.6%
- 4
Gemini 3 Flash ↑3 64.4%
- 5 GPT-6.1 Sol ↓2 64.4%
- 6 GPT-6 Astra ↓1 54.8%
- 7 GPT-6.1 Terra ↓1 46.6%
- 8 Kimi K2.5 35.6%
- 9 Claude Haiku 5.5 23.3%
- 10 GPT-6 Luna ↑1 21.9%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official EarthRecord leaderboard
Claude Opus 5.5 on the official leaderboard passed 85% in September 2026
Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official EarthRecord leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Science and reasoning
- GPQA Diamond In use 198 graduate-level 'Google-proof' multiple-choice questions in biology, chemistry and physics.
Gemini 4 Carbon 97%
- Terminal-Bench-Science In use 70 scientific research-workflow tasks across five science areas, run in Harbor containers. Hidden tests give the resolution rate. GPT-6 Astra 68.1%
- BioMysteryBench In use Bioinformatics problems on anonymized real datasets that require real analysis with tools. Rubric-graded, with human-solvable and human-difficult splits. Claude Fable 5.2 83.7%
- FrontierScience In use Models solve physics, chemistry, and biology problems in an Olympiad track graded by short answer and a Research track graded by 10-point rubrics.
Intern-S2-397B 87%
- HiPhO In use Models solve theory problems from 2024 and 2025 physics olympiad exams, some with diagrams. Official marking schemes grade answers and steps, mapped to medal levels.
Gemini 3 Pro 78.3%
- LABBench2 In use 11 biology research sub-tasks (literature, figures, tables, patents, trials and more) with tool access. Macro-averaged accuracy. GPT-5.2 60.9%
Data audit
All 10 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
- Pass Independent runs Scores from the official leaderboard (12 entries).
- Pass Models scored 11 models we track, plus 1 other entry on its leaderboard.
- Pass Newest score The newest is from October 2026, for GPT-6.1 Terra.
- Pass Frontier models 5 of the top 10 have a score, led by GPT-6 Astra (#4).
- Note Headroom Saturated. The best is 85.4%, past 85% since September 2026.
Leaderboard import 3 pass
- Pass Importer We read the board on ranranrannervous.github.io: 11 rows for models we track and 1 other entry at the last read.
- Pass Last read November 18th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.