RankingEffort benchmarks

Gemma 4 26B-A4B IT

Compare models

29 measured results · 3 effort settings · 19 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

AGemma 4 26B-A4B IT · Reasoning3 of 7 capabilities supported

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

Agentic
A9.2Supported
Hard reasoning
A9.0Supported
Coding
A8.0Preliminary
Human pref
Unknown
Knowledge
A12.7Supported
Multimodal
A11.5Preliminary
Long context
A18.6Preliminary

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

Gemma 4 26B-A4B IT · Reasoning: 3.8 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?

GDPval · 9.2% direct comparison weight
Tau Bench · 9.1% direct comparison weight
briefcase · 8.7% direct comparison weight
AutomationBench · 8.7% direct comparison weight
Terminal-Bench · 8.2% direct comparison weight
Other families (8) · 56.0% 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 2026-10-23.

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	268 → 268	No movement	444 remaining / 0 lost	5.0 (+1.2 points)
Datacurve	268 → 268	No movement	444 remaining / 0 lost	3.0 (-0.8 points)
LiveCodeBench	268 → 268	No movement	444 remaining / 0 lost	3.8 (-0.0 points)
Scale	268 → 268	No movement	444 remaining / 0 lost	4.5 (+0.7 points)
Terminal-Bench	268 → 268	No movement	444 remaining / 0 lost	3.8 (-0.0 points)
Cognition	268 → 268	No movement	444 remaining / 0 lost	3.7 (-0.1 points)
LiveBench	267 → 267	No movement	443 remaining / 1 lost	4.0 (+0.2 points)
SWE-rebench	268 → 267	Up 1	444 remaining / 0 lost	4.0 (+0.1 points)
Epoch AI	259 → 255	Up 4	431 remaining / 13 lost	4.0 (+0.1 points)
Vals AI	267 → 267	No movement	443 remaining / 1 lost	3.7 (-0.2 points)
Provider reports · Anthropic	268 → 268	No movement	444 remaining / 0 lost	3.8 (-0.0 points)
Provider reports · Moonshot AI	268 → 268	No movement	444 remaining / 0 lost	3.7 (-0.1 points)
Provider reports · Meta	268 → 268	No movement	444 remaining / 0 lost	3.8 (-0.0 points)
Provider reports · Mistral	268 → 268	No movement	444 remaining / 0 lost	3.8 (+0.0 points)
Provider reports · OpenAI	268 → 268	No movement	444 remaining / 0 lost	3.9 (+0.0 points)
All provider reports	268 → 268	No movement	444 remaining / 0 lost	3.7 (-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	4.8	+1.0
deepswe	3.0	-0.8
ifbench	3.3	-0.6
critpt	4.4	+0.5
epoch-game-puzzles	3.6	-0.2
aa-lcr	3.6	-0.2
mmmu-pro	4.1	+0.2
simpleqa	4.0	+0.2
gpqa	3.7	-0.2
hle	4.0	+0.2
mmlu-pro	4.0	+0.1
briefcase	4.0	+0.1
frontiercode	3.7	-0.1
swe-rebench	4.0	+0.1
livebench-instructions	3.9	+0.1
gdp-pdf	3.9	+0.1
multi-swe-bench	3.8	-0.1
tau-bench	3.8	-0.1
terminal-bench	3.9	+0.1
omniscience	3.8	-0.1
livebench-language	3.9	+0.1
apex-agents	3.9	+0.1
arc-agi-3	3.8	-0.1
automationbench	3.9	+0.1
terminal-bench-science	3.9	+0.1
enigma-eval	3.8	-0.1
harvey	3.8	-0.0
livebench-coding	3.8	-0.0
swe-atlas-qna	3.8	-0.0
gmmlu	3.9	+0.0
analyst-agent	3.9	+0.0
livebench-reasoning	3.8	-0.0
vibe-code	3.8	-0.0
math500	3.9	+0.0
itbench	3.8	-0.0
swe-atlas-test-writing	3.8	-0.0
vals-code-migration	3.8	-0.0
vals-finance-agent	3.8	-0.0
vals-legal-research	3.8	-0.0
vals-tax-agent	3.8	-0.0
livecodebench	3.9	+0.0
vals-excel-modeling	3.8	-0.0
enterprise-ops	3.8	-0.0
aime	3.8	-0.0
frontiermath	3.8	+0.0
gdpval	3.8	-0.0
livebench-math	3.8	+0.0
swe-atlas-refactoring	3.8	-0.0
swe-bench-pro	3.8	+0.0
livebench-data	3.8	-0.0

What evidence is missing from the fit?

Across all collected settings for this model: 29 observations, 26 contributing, 0 matched without graph weight and 3 excluded. Counts do not establish rank eligibility.

2 · Benchmark family or source protocol has not been reviewed
1 · 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

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.

Benchmark / source	Minimal	Reasoning	Non-reasoning
AA-Briefcase EloArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-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 2026-10-18.	Not reported	315.08 Elo Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	Not reported
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.	Not reported	65.67% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	42.33% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
AutomationBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share 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 2026-10-18.	Not reported	1.94% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	Not reported
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.	Not reported	0% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	0% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
GDPval-AA v2.1 LeaderboardArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Elo 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 2026-10-18.	Not reported	565.75 Elo Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	Not reported
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.	Not reported	79.19% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	71.41% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
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.	Not reported	19.32% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	11.49% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
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.	Not reported	72.45% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	45.44% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
ITBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Combined public and private Site Reliability Engineering (SRE) tasks · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	Not reported	23.63% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	Not reported
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.	Not reported	69.25% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	66.71% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
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.	Not reported	-50.83 Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	-62.32 Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
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.	Not reported	19.1% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	15.48% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
SciCodeArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark 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 2026-10-18.	Not reported	40.05% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	Not reported
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.	Not reported	43.57% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	40.35% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
𝜏³-BankingArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by Sierra Research · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	Not reported	11.96% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	Not reported
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.	Not reported	13.64% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	25% Reported settings Gemma 4 26B A4B (Non-reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b-non-reasoning
Terminal-Bench v2.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by the Laude Institute and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	Not reported	38.95% Reported settings Gemma 4 26B A4B (Reasoning); Model source: https://artificialanalysis.ai/models/gemma-4-26b-a4b	Not reported
Epoch · GPQA diamond · v1.0.11Epoch AI · Epoch AI Inspect; task 1.0.11; scorer choice Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. Unit: percent. Source collected 2026-10-18.	73.23% Reported settings Epoch AI Inspect; task 1.0.11; scorer choice; model=gemma-4-26b-a4b-it_minimal; run=Usx9KaiXvHZU4qrzgc4sLg; mean accuracy; https://epoch.ai/data/benchmarks.csv	Not reported	Not reported
Epoch · OTIS Mock AIME 2024-2025 · v1.1.6Epoch AI · Epoch AI Inspect; task 1.1.6; scorer model_extracted_exact_match Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. Unit: percent. Source collected 2026-10-18.	82.22% Reported settings Epoch AI Inspect; task 1.1.6; scorer model_extracted_exact_match; model=gemma-4-26b-a4b-it_minimal; run=mstwSH7noJfvhGcd6Lz5kd; mean accuracy; https://epoch.ai/data/benchmarks.csv	Not reported	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 · 8,916 results across 259 models · Collected 2026-10-18

Gemma 4 26B-A4B IT

Compare with

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context

Capability score evidence

Benchmark results