Model
Gemini 3.8 Flash
Knowledge & Reasoning
Knowledge & Reasoning| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| ArcticAbstain | 8.4 | abstention_rate | N/A | Independent | medium | arxiv.org |
| CritPt | 18.3 | accuracy | N/A | Independent | low | epoch.ai |
| GPQA Diamond | 95.4 | accuracy | N/A | Independent | high | epoch.ai |
| HLE | 44.5 | accuracy | N/A | Independent | low | epoch.ai |
| InsClaimBench | 79.3 | accuracy | N/A | Independent | medium | arxiv.org |
| SimpleQA Verified | 69.7 | accuracy | N/A | Independent | high | epoch.ai |
Coding
Coding| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| DeepSWE | 73.8 | pass@1 | mini-swe-agent | Independent | medium | deepswe.datacurve.ai |
| PLCWorld | 32.1 | success_rate | N/A | Independent | medium | arxiv.org |
| SciCode | 56.6 | score | N/A | Independent | medium | artificialanalysis.ai |
| TestSuiteBench | 37.9 | score | N/A | Independent | medium | arxiv.org |
| Video2World | 30.1 | v2wscore | N/A | Independent | medium | arxiv.org |
Agentic / Tool Use
Agentic / Tool Use| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| Argo-Bench | 26.2 | score | N/A | Independent | medium | arxiv.org |
| BioStudyBench | 94.7 | pass_rate | N/A | Independent | medium | arxiv.org |
| BrickBench | 1014.0 | elo | N/A | Independent | medium | arxiv.org |
| CompMat-Bench | 89.4 | pass_rate | matclaw | Independent | medium | arxiv.org |
| cua-speedrun | 91.6 | success_rate | N/A | Independent | medium | arxiv.org |
| KlinikeBench | 18.9 | strict_pass@1 | N/A | Independent | medium | arxiv.org |
| Mine Odyssey | 26.7 | success_rate | N/A | Independent | medium | arxiv.org |
| MMPostTrainBench | 42.9 | mean_task_score_(%) | gemini cli | Independent | medium | arxiv.org |
| RoboQuest | 2.0 | success_rate | N/A | Independent | medium | arxiv.org |
| RobotWorld | 1.2 | success_rate | N/A | Independent | medium | arxiv.org |
| StoreBench | 0.6 | composite | N/A | Independent | medium | arxiv.org |
| TALK-Dem | 97.0 | success_rate | N/A | Independent | medium | arxiv.org |
| UndoBench | 21.1 | success_rate | N/A | Independent | medium | arxiv.org |
| WorldAuditBench | 32.4 | success_rate | vlm auditors | Independent | medium | arxiv.org |
| WorldAuditBench | 17.4 | success_rate | vla–vlm auditors | Independent | medium | arxiv.org |
Multimodal
Multimodal| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| 4MT-VLM | 44.0 | accuracy | N/A | Independent | medium | arxiv.org |
| EgoGears | 79.8 | accuracy | N/A | Independent | medium | arxiv.org |
| EMBER-Bench | 72.8 | accuracy | N/A | Independent | medium | arxiv.org |
| GDP.pdf | 23.4 | gdp.pdf_score | N/A | Independent | medium | surgehq.ai |
| HeiCo-FOCUS | 51.4 | accuracy | N/A | Independent | medium | arxiv.org |
| HSS | 41.6 | pass@1 | N/A | Independent | medium | arxiv.org |
| PlotGround-1k | 65.5 | accuracy | N/A | Independent | medium | arxiv.org |
Long Context
Long Context| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| LongHarness Bench | 10.0 | accuracy | direct | Independent | medium | arxiv.org |
Instruction Following
Instruction Following| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| Design Creativity Bench | 60.6 | overall | N/A | Independent | medium | arxiv.org |
Safety & Alignment
Safety & Alignment| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| APEX-Harm | 11.8 | success_rate | N/A | Independent | medium | arxiv.org |
| DRYLAB-Bench | 0.4 | floor-included_brs_s2 | N/A | Independent | medium | arxiv.org |
| FIGS | 65.5 | automated_judge_applying_a_10-rule_taxonomy_separating_sycophancy_from_calibrated_validation | N/A | Independent | medium | arxiv.org |
| Nonpublic Vulnerabilities Benchmark | 0.0 | attack_score | opencode | Independent | medium | arxiv.org |
Abstract Reasoning
Abstract ReasoningMultilingual
Multilingual| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| HakemBench | 0.9 | composite | N/A | Independent | medium | arxiv.org |
Reproduced by the community
Loading community runs…