RankingEffort benchmarks
Gemini 4 Argon
58 measured results · 4 effort settings · 36 benchmark/harness combinations
Change model
Capability profile
One configuration, seven capabilities
- AGemini 4 Argon · High3 of 7 capabilities supported
Gaps are unknown, not zero. Hollow points are preliminary.
Agentic
A83.0SupportedHard reasoning
A75.9SupportedCoding
A88.0SupportedHuman pref
UnknownKnowledge
A68.3PreliminaryMultimodal
UnknownLong context
A50.1Preliminary
Each profile belongs to the named configuration. Capability scores use a shared panel for all axes, distinct from the overall ranking panel; they are not accuracy percentages or direct head-to-head wins.
Capability score evidence
Gemini 4 Argon · High: 88.3 overall comparison score. Meets the overall evidence requirements.
Shared-result disagreement. This aggregate order conflicts with matched benchmark results against 1 other configurations. Different test coverage and opponents affect the fit. Inspect published results before treating the rank as conclusive.
Which sources carry the weight?
- Artificial Analysis · 23.6% direct comparison weight · independent of the model provider
- ARC Prize · 22.4% direct comparison weight · independent of the model provider
- Vals AI · 20.4% direct comparison weight · independent of the model provider
- Scale · 18.0% direct comparison weight · independent of the model provider
- Cognition · 15.6% direct comparison weight · independent of the model provider
Which benchmark families carry the weight?
- SWE-bench Pro · 18.0% direct comparison weight
- FrontierCode · 15.6% direct comparison weight
- ARC-AGI · 12.8% direct comparison weight
- arc-agi-3 · 9.6% direct comparison weight
- Terminal-Bench · 4.5% direct comparison weight
- Other families (17) · 39.5% direct comparison weight
These shares describe direct comparison weight, not fractions of the final score or confidence. Source balancing does not give every family equal weight or remove correlations between overlapping sources.
How sensitive is this score?
The audit removes one source at a time and refits against the same overall reference panel. These published results use the matching dataset audited 2026-11-22.
Both ranks use the same surviving configurations for each removal. The cohort starts with 602 configurations supported in the original audit, including separate effort settings. These are diagnostic positions, not the default leaderboard ranks. Eligibility is not rechecked after removal.
| Removed source | Rank before → after | Rank movement | Cohort remaining / lost support | Score after (change) |
|---|---|---|---|---|
| Artificial Analysis | 14 → 10 | Up 4 | 230 remaining / 372 lost | 91.1 |
| ARC Prize | 16 → 13 | Up 3 | 602 remaining / 0 lost | 90.4 |
| Datacurve | 16 → 17 | Down 1 | 602 remaining / 0 lost | 86.7 |
| LiveCodeBench | 16 → 16 | No movement | 602 remaining / 0 lost | 88.3 |
| Scale | 16 → 22 | Down 6 | 602 remaining / 0 lost | 86.0 |
| Terminal-Bench | 16 → 15 | Up 1 | 602 remaining / 0 lost | 88.5 |
| Cognition | 16 → 17 | Down 1 | 602 remaining / 0 lost | 85.8 |
| LiveBench | 16 → 16 | No movement | 601 remaining / 1 lost | 89.2 |
| SWE-rebench | 16 → 16 | No movement | 602 remaining / 0 lost | 88.1 |
| Epoch AI | 16 → 16 | No movement | 588 remaining / 14 lost | 88.7 |
| Vals AI | 16 → 15 | Up 1 | 601 remaining / 1 lost | 88.5 |
| Provider reports · Anthropic | 16 → 15 | Up 1 | 602 remaining / 0 lost | 88.2 |
| Provider reports · Moonshot AI | 16 → 16 | No movement | 602 remaining / 0 lost | 87.9 |
| Provider reports · Meta | 16 → 16 | No movement | 602 remaining / 0 lost | 88.2 |
| Provider reports · Mistral | 16 → 16 | No movement | 602 remaining / 0 lost | 88.3 |
| Provider reports · OpenAI | 16 → 14 | Up 2 | 602 remaining / 0 lost | 88.2 |
| All provider reports | 16 → 13 | Up 3 | 602 remaining / 0 lost | 87.6 |
Lost support means a configuration no longer has a comparison path to the complete reference panel. Its rank is unavailable. A surviving path alone does not establish leaderboard eligibility.
Overlapping sources may remain correlated even with source balancing. Leave-one-source-out movement is a sensitivity diagnostic, not a spread of source-specific ranks or a confidence measure. Scoring methodology · Dated configuration audit
Every benchmark family
| Removed evidence | Recomputed score | Change |
|---|---|---|
| swe-bench-pro | 85.5 | -2.8 |
| frontiercode | 85.8 | -2.5 |
| arc-agi | 90.5 | +2.2 |
| terminal-bench | 90.3 | +2.0 |
| deepswe | 86.7 | -1.6 |
| arc-agi-3 | 87.6 | -0.7 |
| aa-lcr | 88.8 | +0.5 |
| vals-tax-agent | 87.8 | -0.5 |
| hle | 88.8 | +0.5 |
| vals-finance-agent | 87.9 | -0.4 |
| automationbench | 87.9 | -0.4 |
| vibe-code | 87.9 | -0.4 |
| harvey | 88.6 | +0.3 |
| vals-legal-research | 88.0 | -0.3 |
| vals-code-migration | 88.0 | -0.3 |
| mmmu-pro | 88.6 | +0.3 |
| vals-excel-modeling | 88.0 | -0.3 |
| enigma-eval | 88.5 | +0.2 |
| simpleqa | 88.5 | +0.2 |
| gdp-pdf | 88.5 | +0.2 |
| terminal-bench-science | 88.1 | -0.2 |
| swe-rebench | 88.1 | -0.2 |
| ifbench | 88.1 | -0.2 |
| epoch-game-puzzles | 88.1 | -0.2 |
| omniscience | 88.5 | +0.2 |
| tau-bench | 88.1 | -0.2 |
| swe-atlas-test-writing | 88.1 | -0.1 |
| gpqa | 88.2 | -0.1 |
| swe-atlas-qna | 88.2 | -0.1 |
| mmlu-pro | 88.4 | +0.1 |
| livebench-data | 88.4 | +0.1 |
| analyst-agent | 88.4 | +0.1 |
| enterprise-ops | 88.4 | +0.1 |
| gdpval | 88.2 | -0.1 |
| multi-swe-bench | 88.2 | -0.1 |
| livebench-instructions | 88.4 | +0.1 |
| frontiermath | 88.2 | -0.1 |
| livebench-coding | 88.2 | -0.1 |
| livebench-language | 88.3 | +0.1 |
| livebench-math | 88.2 | -0.0 |
| livebench-reasoning | 88.3 | -0.0 |
| briefcase | 88.3 | -0.0 |
| swe-atlas-refactoring | 88.3 | -0.0 |
| apex-agents | 88.3 | +0.0 |
| itbench | 88.3 | -0.0 |
| aime | 88.3 | -0.0 |
| livecodebench | 88.3 | -0.0 |
| critpt | 88.3 | +0.0 |
| gmmlu | 88.3 | -0.0 |
| math500 | 88.3 | -0.0 |
What evidence is missing from the fit?
Across all collected settings for this model: 90 observations, 46 contributing, 0 matched without graph weight and 44 excluded. Counts do not establish rank eligibility.
- 2 · Benchmark family or source protocol has not been reviewed
- 2 · Original SciCode grading defects are under review; retained as supporting evidence, excluded from aggregate and capability scores. Corrected SciCode-Verified requires a separate protocol review. Sources: https://arxiv.org/abs/2608.04975 and https://artificialanalysis.ai/evaluations/scicode
- 5 · Evaluator has not been reviewed for independent admission
- 1 · Cache hit rate is much lower due to a non-production setting.
- 2 · Effort is not identified
- 18 · Provider-published claim. Independent source results are admitted separately with exact checkpoint, effort and compatible protocol; this grid does not establish a matched comparison.
- 1 · Provider-published claim. Independent source results are admitted separately with exact checkpoint, effort and compatible protocol; this grid does not establish a matched comparison. Historical provider claim; secondary and superseded for current FrontierSWE comparisons by the independent Proximal FrontierSWE v2 mean@5 board at https://www.frontierswe.com/. Version-unspecified claims are not v2 results. Excluded from weighted Overall.
- 3 · No reviewed effort for this catalog observation; any separately collected effort measurement is counted separately
- 9 · Supporting evidence outside the reviewed capability core
- 1 · No matched opponent in this evaluation unit
Benchmark results
Reviewed retained observations are in use. October 2026 Real-SWE board has no 95% confidence whisker, so these scores stay at the 2026-10-18 retrieval. Original source retrieval 2026-10-18; source updated 2026-10-18.
| Benchmark / source | Low | Medium | High | Unspecified |
|---|---|---|---|---|
AA-AnalystAgentBenchmark detailsShare of tasks solved on all 5 of 5 attempts · 80 tasks, 5 attempts per task · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | Not reported | 48.75% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
AA-Briefcase EloBenchmark detailsAA-Briefcase v1.1 is an agentic knowledge work benchmark developed by Artificial Analysis. AA-Briefcase Elo is a combined metric that aggregates rubric pass rate, analytical quality Elo and presentation Elo · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-11-22. | Not reported | 1,466.63 Elo Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 1,485.08 Elo Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
AA-LCR v1.1Benchmark detailsIndependently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 78.33% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 79.67% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
AutomationBench-AABenchmark detailsShare of task objectives completed with no guardrail violations · Higher is better · Benchmark developed by Zapier · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 75.82% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 77.51% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
CritPtBenchmark detailsBenchmark developed by Argonne and UIUC, with contributions from 60+ researchers globally · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 23.14% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 27.14% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
GDP.pdf: All-pass RateBenchmark detailsShare of attempts where every atomic criterion passed · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 25.6% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 21.8% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
GDPval-AA v2.1 LeaderboardBenchmark detailsElo rating for performance on real-world work tasks · Anchored to DeepSeek V4.1 Flash (max) at 1600 · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-11-22. | Not reported | 1,578.39 Elo Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 1,633.51 Elo Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
Harvey LAB-AA v1.1: Hallucination-Gated All-Pass RateBenchmark detailsAverage over tasks of the share of the judge panel finding every rubric criterion passed, with a task zeroed on any material hallucination · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | Not reported | 3.89% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
Humanity's Last ExamBenchmark detailsIndependently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 49.91% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 57.09% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
MLCR-AABenchmark detailsShare of responses judged complete with no claim contradicting the record · Expert + compound (hard) sets · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | Not reported | 13.89% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
MMMU-ProBenchmark detailsIndependently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 88.32% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | Not reported | Not reported |
AA-Omniscience IndexBenchmark detailsAA-Omniscience v1.1 Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. · Evaluation results measured independently by Artificial Analysis Unit: index. Source collected 2026-11-22. | Not reported | 41.1 Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 46.1 Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
AA-Omniscience AccuracyBenchmark detailsAA-Omniscience v1.1 Accuracy (higher is better) measures the proportion of correctly answered questions out of all questions, regardless of whether the model chooses to answer · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 48.97% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 48.3% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
SciCodeBenchmark detailsBenchmark curated by scientists across 16 disciplines. SciCode is under review following an independent audit of the dataset · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 58.91% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 61.81% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
Terminal-Bench-Science 0.1Benchmark detailsShare of tasks passed (pass@1), averaged over three repeats · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 48.57% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 52.86% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
Terminal-Bench v4.0Benchmark detailsBenchmark developed by the Laude Institute, Stanford, and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-11-22. | Not reported | 55.05% Reported settingsGemini 4 Argon (Medium); Model source: https://artificialanalysis.ai/models/gemini-4-argon-medium | 57.07% Reported settingsGemini 4 Argon (High); Model source: https://artificialanalysis.ai/models/gemini-4-argon | Not reported |
ARC-AGI-2Benchmark detailsOriginal public board. Configuration and task coverage may differ; inspect source. Unit: percent. Source collected 2026-11-22. | 42.08% Reported settingsGemini 4 Argon (Low); Source model ID: google-gemini-4-argon-low; 120 tasks | 87.08% Reported settingsGemini 4 Argon (Medium); Source model ID: google-gemini-4-argon-medium; 120 tasks | 91.25% Reported settingsGemini 4 Argon (High); Source model ID: google-gemini-4-argon-high; 120 tasks | Not reported |
SWE-Bench Pro V2 Hard · mini-swe-agentBenchmark details51-task Hard series run with mini-swe-agent. Compare models only within this agent. Unit: percent. Source collected 2026-11-22. | Not reported | Not reported | 92.16% Reported settingsGemini 4 Argon; Reported effort: high; 51 tasks; agent mini-swe-agent; source date 2026-10-14 | Not reported |
ARC-AGI-3 · provider-adapterBenchmark detailsARC-AGI-3; Provider Adapter. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-11-22. | 97.46% Reported settingsARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group google-gemini-4-argon; variant provider-adapter-low; public-demo task scores are excluded | 97.34% Reported settingsARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group google-gemini-4-argon; variant provider-adapter-medium; public-demo task scores are excluded | 97.3% Reported settingsARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group google-gemini-4-argon; variant provider-adapter-high; public-demo task scores are excluded | Not reported |
ARC-AGI-3 · standardBenchmark detailsARC-AGI-3; Standard. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-11-22. | 49.79% Reported settingsARC-AGI-3 Semi-Private verified summary; Standard harness; source group google-gemini-4-argon; variant low; public-demo task scores are excluded | 46.27% Reported settingsARC-AGI-3 Semi-Private verified summary; Standard harness; source group google-gemini-4-argon; variant medium; public-demo task scores are excluded | Not reported | Not reported |
FrontierCode 1.1 · extended-devin-cliBenchmark detailsFrontierCode 1.1; devin-cli. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-11-22. | 60.58% Reported settingsFrontierCode 1.1; extended; 150 tasks; devin-cli; source effort low; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0229 | 60.5% Reported settingsFrontierCode 1.1; extended; 150 tasks; devin-cli; source effort medium; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0134 | 59.38% Reported settingsFrontierCode 1.1; extended; 150 tasks; devin-cli; source effort high; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0137 | Not reported |
FrontierCode 1.1 · main-devin-cliBenchmark detailsFrontierCode 1.1; devin-cli. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-11-22. | 47.2% Reported settingsFrontierCode 1.1; main; 100 tasks; devin-cli; source effort low; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0304 | 47.55% Reported settingsFrontierCode 1.1; main; 100 tasks; devin-cli; source effort medium; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0121 | 45.87% Reported settingsFrontierCode 1.1; main; 100 tasks; devin-cli; source effort high; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0103 | Not reported |
Vals AI · Finance Agent v2Benchmark detailsOverall weighted rubric accuracy, averaged over three runs on 450 tasks. Dealbreaker failures and timeouts score zero. Related categories and all-pass scores are not additional votes. Three-model judge panel; private test set. Unit: percent. Source collected 2026-11-22. | Not reported | Not reported | 65.4% Reported settings{"checkpoint":"gemini-4-argon","evaluatorModel":"google/gemini-4-argon","sourceUpdatedOn":"2026-10-29","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":1,"top_p":null,"max_output_tokens":262144,"verbosity":null,"provider":"Google","harness":null} | Not reported |
Vals AI · Code Migration v1Benchmark detailsHidden behavior-test pass rate, equally weighted across 40 repositories (30 CLI and 10 COBOL-to-Java). Anti-cheat failures score zero. Code-quality ratings and language splits are not additional votes. Unit: percent. Source collected 2026-11-22. | Not reported | Not reported | 68.17% Reported settings{"checkpoint":"gemini-4-argon","evaluatorModel":"google/gemini-4-argon","sourceUpdatedOn":"2026-10-29","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":1,"top_p":null,"max_output_tokens":262144,"verbosity":null,"provider":"Google","harness":null} | Not reported |
Vals AI · Legal Research Bench v1Benchmark detailsAll-pass accuracy on 208 private legal research tasks. Every expert rubric check must pass; GPT-5.4 judges responses. Partial-credit scores and practice-area splits are not additional votes. Unit: percent. Source collected 2026-11-22. | Not reported | Not reported | 54.81% Reported settings{"checkpoint":"gemini-4-argon","evaluatorModel":"google/gemini-4-argon","sourceUpdatedOn":"2026-11-19","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":1,"top_p":null,"max_output_tokens":262144,"verbosity":null,"provider":"Google","harness":null} | Not reported |
What these comparisons mean
Each source reports its own effort labels. Matching labels do not establish equal compute budgets or identical fallback behavior. Open Reported settings to inspect the exact model name, agent and task coverage. Multiple results at one level remain visible rather than selecting the best. Source-reported None or Non-reasoning labels are not treated as evidence that an API supports those settings.
These measurements use their own dated source collection. The leaderboard fits identified configurations jointly; it contains no pooled model ratings. Explicit thinking budgets are separate configurations. Read the methodology.
Download all effort benchmark results · 10,184 results across 381 models · Collected 2026-11-22