RankingEffort benchmarks

Gemini 2.5 Flash · Reasoning

Compare models

30 measured results · 2 effort settings · 15 benchmark/harness combinations

Model evidence and sources →

Change model

Model

Current configurationSources

Compare with

Choose a second model and its effort to see both profiles and benchmark differences below.

Compare withModel and configuration

Capability profile

One configuration, seven capabilities

Comparison score · 0–100

AGemini 2.5 Flash · Reasoning3 of 7 capabilities supported

Gaps are unknown, not zero. Hollow points are preliminary.

Agentic
A3.1Preliminary
Hard reasoning
A12.3Supported
Coding
A6.4Supported
Human pref
Unknown
Knowledge
A12.0Supported
Multimodal
A10.7Preliminary
Long context
A20.6Preliminary

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 2.5 Flash · Reasoning: 4.1 overall comparison score. Meets the overall evidence requirements.

6 areas measured. The available task mix affects this estimate; missing areas remain unknown.

Which sources carry the weight?

Artificial Analysis · 100.0% direct comparison weight · independent of the model provider

Which benchmark families carry the weight?

MMLU-Pro · 8.2% direct comparison weight
IFBench · 8.1% direct comparison weight
LiveCodeBench · 8.0% direct comparison weight
GPQA · 8.0% direct comparison weight
Humanity’s Last Exam · 8.0% direct comparison weight
Other families (8) · 59.8% 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 2026-10-20.

Both ranks use the same surviving configurations for each removal. The cohort starts with 444 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.

What changes when a source is removed?
Removed source	Rank before → after	Rank movement	Cohort remaining / lost support	Score after (change)
Artificial Analysis	Comparison support lost; no rank		203 remaining / 241 lost	Unavailable
ARC Prize	263 → 264	Down 1	444 remaining / 0 lost	5.4 (+1.3 points)
Datacurve	263 → 263	No movement	444 remaining / 0 lost	3.3 (-0.9 points)
LiveCodeBench	263 → 263	No movement	444 remaining / 0 lost	4.1 (-0.1 points)
Scale	263 → 263	No movement	444 remaining / 0 lost	4.9 (+0.7 points)
Terminal-Bench	263 → 263	No movement	444 remaining / 0 lost	4.1 (-0.0 points)
Cognition	263 → 264	Down 1	444 remaining / 0 lost	4.0 (-0.1 points)
LiveBench	262 → 262	No movement	443 remaining / 1 lost	4.3 (+0.2 points)
SWE-rebench	263 → 263	No movement	444 remaining / 0 lost	4.3 (+0.1 points)
Epoch AI	254 → 252	Up 2	431 remaining / 13 lost	4.2 (+0.0 points)
Vals AI	263 → 263	No movement	443 remaining / 1 lost	4.0 (-0.2 points)
Provider reports · Anthropic	263 → 263	No movement	444 remaining / 0 lost	4.1 (-0.0 points)
Provider reports · Moonshot AI	263 → 263	No movement	444 remaining / 0 lost	4.0 (-0.1 points)
Provider reports · Meta	263 → 263	No movement	444 remaining / 0 lost	4.1 (-0.0 points)
Provider reports · Mistral	263 → 263	No movement	444 remaining / 0 lost	4.1 (+0.0 points)
Provider reports · OpenAI	263 → 263	No movement	444 remaining / 0 lost	4.2 (+0.0 points)
All provider reports	263 → 263	No movement	444 remaining / 0 lost	4.0 (-0.1 points)

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
arc-agi	5.2	+1.1
deepswe	3.3	-0.9
omniscience	3.5	-0.6
hle	4.6	+0.5
tau-bench	4.6	+0.4
critpt	3.8	-0.3
mmmu-pro	4.4	+0.3
terminal-bench	4.4	+0.3
epoch-game-puzzles	3.9	-0.3
aa-lcr	3.9	-0.2
simpleqa	4.4	+0.2
gpqa	4.0	-0.1
frontiercode	4.0	-0.1
aime	4.3	+0.1
livecodebench	4.3	+0.1
swe-rebench	4.3	+0.1
gdp-pdf	4.3	+0.1
livebench-instructions	4.2	+0.1
multi-swe-bench	4.1	-0.1
math500	4.1	-0.1
arc-agi-3	4.1	-0.1
livebench-language	4.2	+0.1
terminal-bench-science	4.2	+0.1
apex-agents	4.2	+0.1
enigma-eval	4.1	-0.1
livebench-coding	4.1	-0.0
swe-atlas-qna	4.1	-0.0
harvey	4.1	-0.0
ifbench	4.1	-0.0
analyst-agent	4.2	+0.0
itbench	4.1	-0.0
livebench-reasoning	4.1	-0.0
vibe-code	4.1	-0.0
gmmlu	4.2	+0.0
automationbench	4.1	-0.0
swe-atlas-test-writing	4.1	-0.0
briefcase	4.1	-0.0
vals-code-migration	4.1	-0.0
vals-finance-agent	4.1	-0.0
vals-legal-research	4.1	-0.0
gdpval	4.2	+0.0
vals-tax-agent	4.1	-0.0
mmlu-pro	4.2	+0.0
vals-excel-modeling	4.1	-0.0
frontiermath	4.2	+0.0
livebench-math	4.2	+0.0
swe-atlas-refactoring	4.1	-0.0
enterprise-ops	4.1	+0.0
swe-bench-pro	4.1	+0.0
livebench-data	4.1	-0.0

What evidence is missing from the fit?

Across all collected settings for this model: 52 observations, 28 contributing, 0 matched without graph weight and 24 excluded. Counts do not establish rank eligibility.

2 · Benchmark family or source protocol has not been reviewed
14 · No reviewed effort for this catalog observation; any separately collected effort measurement is counted separately
8 · Supporting evidence outside the reviewed capability core

Inspect every observation and exclusion →

Benchmark results

These are published benchmark scores. Supported effort estimates also appear on the Capability leaderboard. Results from different harnesses occupy different rows. Missing cells stay unknown; no scores are copied between effort levels.

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 · 8,915 results across 259 models · Collected 2026-10-18

Benchmark / source	Reasoning	Non-reasoning
AIME 2025Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	73.33% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	60.33% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
AA-LCR v1.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	65.33% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	49.9% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
CritPtArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by Argonne and UIUC, with contributions from 60+ researchers globally. CritPt is under review by its authors following external feedback · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	1.14% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	1.43% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
GPQA DiamondArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	78.99% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	68.28% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
Humanity's Last ExamArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	12.14% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	4.73% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
IFBenchArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	50.27% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	38.98% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
LiveCodeBenchArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	69.52% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	49.52% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
MATH-500Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	98.13% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	93.2% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
MLCR-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of tasks judged accurate, complete, and concise · Expert + compound (hard) sets · Benchmark developed by Wisedocs · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	8.89% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	8.33% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
MMLU-ProArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	83.15% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	80.92% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
MMMU-ProArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	69.08% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	65.49% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
AA-Omniscience IndexArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Omniscience 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 2026-10-18.	-29.8 Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	-42.57 Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
AA-Omniscience AccuracyArtificial Analysis · 6000 questions; 1 attempt; correct-answer share; GPT-5.6 Luna medium grader Benchmark details AA-Omniscience 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 2026-10-18.	25.95% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	26.13% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
𝜏²-Bench TelecomArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	31.58% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	14.91% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash
Terminal-Bench HardArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	13.64% Reported settings Gemini 2.5 Flash (Reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash-reasoning	12.12% Reported settings Gemini 2.5 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemini-2-5-flash

Gemini 2.5 Flash · Reasoning

Compare with

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context

Capability score evidence

Benchmark results