RankingEffort benchmarks

Step 5 Preview

Compare models

19 measured results · 1 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

AStep 5 Preview · Reasoning2 of 7 capabilities supported

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

Agentic
A57.0Supported
Hard reasoning
A55.0Supported
Coding
A74.8Preliminary
Human pref
Unknown
Knowledge
A35.0Preliminary
Multimodal
A27.6Preliminary
Long context
A59.8Preliminary

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

Step 5 Preview · Reasoning: 47.8 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?

Humanity’s Last Exam · 8.0% direct comparison weight
Critpt · 8.0% direct comparison weight
omniscience · 8.0% direct comparison weight
Mmmu Pro · 7.9% direct comparison weight
GDPval · 7.4% direct comparison weight
Other families (11) · 60.7% 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 2026-10-08.

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		204 remaining / 240 lost	Unavailable
ARC Prize	66 → 64	Up 2	444 remaining / 0 lost	53.4 (+5.6 points)
Datacurve	66 → 70	Down 4	444 remaining / 0 lost	43.8 (-4.0 points)
LiveCodeBench	66 → 66	No movement	444 remaining / 0 lost	47.7 (-0.1 points)
Scale	66 → 66	No movement	444 remaining / 0 lost	49.6 (+1.8 points)
Terminal-Bench	66 → 66	No movement	444 remaining / 0 lost	48.1 (+0.3 points)
Cognition	66 → 68	Down 2	444 remaining / 0 lost	46.6 (-1.2 points)
LiveBench	66 → 67	Down 1	443 remaining / 1 lost	48.7 (+0.9 points)
SWE-rebench	66 → 65	Up 1	444 remaining / 0 lost	48.0 (+0.2 points)
Epoch AI	65 → 70	Down 5	426 remaining / 18 lost	48.2 (+0.4 points)
Vals AI	66 → 65	Up 1	443 remaining / 1 lost	46.8 (-1.0 points)
Provider reports · Anthropic	66 → 66	No movement	444 remaining / 0 lost	47.8 (-0.0 points)
Provider reports · Moonshot AI	66 → 66	No movement	444 remaining / 0 lost	47.3 (-0.5 points)
Provider reports · Meta	66 → 67	Down 1	444 remaining / 0 lost	47.7 (-0.1 points)
Provider reports · Mistral	66 → 66	No movement	444 remaining / 0 lost	47.8 (+0.0 points)
Provider reports · OpenAI	66 → 67	Down 1	444 remaining / 0 lost	48.0 (+0.2 points)
All provider reports	66 → 68	Down 2	444 remaining / 0 lost	47.4 (-0.4 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	52.6	+4.8
deepswe	43.8	-4.0
mmmu-pro	51.4	+3.6
terminal-bench-science	50.7	+2.9
harvey	45.5	-2.3
itbench	45.6	-2.2
omniscience	49.9	+2.1
gdp-pdf	49.9	+2.1
frontiercode	46.6	-1.2
epoch-game-puzzles	46.9	-0.9
simpleqa	48.6	+0.8
apex-agents	47.0	-0.8
terminal-bench	47.0	-0.8
enterprise-ops	47.1	-0.7
gdpval	47.1	-0.7
ifbench	47.1	-0.7
briefcase	47.2	-0.6
arc-agi-3	47.3	-0.5
automationbench	48.3	+0.5
mmlu-pro	48.2	+0.4
analyst-agent	48.2	+0.4
aa-lcr	47.4	-0.4
multi-swe-bench	47.4	-0.4
livebench-instructions	48.1	+0.3
gpqa	47.5	-0.3
tau-bench	47.5	-0.3
hle	48.1	+0.3
enigma-eval	48.1	+0.3
livebench-language	48.1	+0.3
swe-atlas-qna	47.5	-0.3
livebench-coding	47.6	-0.2
swe-rebench	48.0	+0.2
vibe-code	47.7	-0.1
swe-atlas-test-writing	47.7	-0.1
livebench-reasoning	47.7	-0.1
livecodebench	47.7	-0.1
aime	47.7	-0.1
vals-legal-research	47.7	-0.1
vals-tax-agent	47.7	-0.1
vals-finance-agent	47.7	-0.1
vals-code-migration	47.7	-0.1
swe-atlas-refactoring	47.7	-0.1
frontiermath	47.8	+0.0
critpt	47.8	-0.0
livebench-data	47.8	+0.0
vals-excel-modeling	47.8	-0.0
math500	47.8	+0.0
livebench-math	47.8	+0.0
gmmlu	47.8	+0.0
swe-bench-pro	47.8	+0.0

What evidence is missing from the fit?

Across all collected settings for this model: 23 observations, 17 contributing, 0 matched without graph weight and 6 excluded. Counts do not establish rank eligibility.

1 · 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
2 · Supporting evidence outside the reviewed capability core
2 · No reviewed effort for this catalog observation; any separately collected effort measurement is counted separately

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	Reasoning
AA-AnalystAgentArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share 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 2026-10-08.	35% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	1,425.23 Elo Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
APEX-Agents-AAArtificial 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 2026-10-08.	37.98% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	88.33% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	50.96% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	20.86% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
EnterpriseOps-Gym-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Overall task success rate across all eight enterprise domains (oracle tool mode) · Benchmark developed by ServiceNow Research · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-08.	47.18% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
GDP.pdf: All-pass RateArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of attempts where every atomic criterion passed · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-08.	14.8% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	1,584.24 Elo Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
Harvey LAB-AA: Criterion Pass RateArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of atomic pass/fail rubric criteria the deliverables satisfy, graded by a single LLM judge · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-08.	93.44% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	46.48% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	55.56% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	16.67% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	76.36% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	16.38 Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	41.53% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
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 2026-10-08.	58.91% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
Terminal-Bench-Science 0.1Artificial Analysis · mini-swe-agent; Terminal-Bench-Science 0.1.0; 70 tasks; pass@1 averaged over 3 repeats Benchmark details Share 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 2026-10-08.	2.38% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified
Terminal-Bench v4.0Artificial Analysis · mini-SWE-agent v2.4.6; 66 tasks; pass@1 averaged over 3 repeats; 500 steps; task timeout up to 8 hours Benchmark details Benchmark 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 2026-10-08.	33.33% Reported settings Step 5 Preview; Reported effort: reasoning; Model source: https://artificialanalysis.ai/models/step-5; Mode from evaluator isReasoning metadata; effort budget not specified

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 · 8,798 results across 256 models · Collected 2026-10-08

Step 5 Preview

Compare with

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context

Capability score evidence

Benchmark results