RankingEffort benchmarks

GPT-6.2 Sol

Compare models

55 measured results · 5 effort settings · 11 benchmark/harness combinations

Change model

Compare with

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

Compare with

Capability profile

One configuration, seven capabilities

Comparison score · 0–100
GPT-6.2 Sol · Max capability radarComparison scores from 0 to 100. Unknown capabilities have no point; lines stop at gaps. Hollow points are preliminary. Exact values and support labels follow the chart.AgenticHard reasoningCodingHuman prefKnowledgeMultimodalLong context50100GPT-6.2 Sol · Max · Agentic: 81.3 · SupportedGPT-6.2 Sol · Max · Hard reasoning: 69.4 · PreliminaryGPT-6.2 Sol · Max · Coding: 92.2 · PreliminaryGPT-6.2 Sol · Max · Knowledge: 86.4 · PreliminaryGPT-6.2 Sol · Max · Long context: 90.6 · Preliminary

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

  • Agentic

    A81.3Supported
  • Hard reasoning

    A69.4Preliminary
  • Coding

    A92.2Preliminary
  • Human pref

    Unknown
  • Knowledge

    A86.4Preliminary
  • Multimodal

    Unknown
  • Long context

    A90.6Preliminary

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

GPT-6.2 Sol · Max: 79.6 overall comparison score. Meets the overall evidence requirements.

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

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?
  • ARC Prize · 67.2% direct comparison weight · independent of the model provider
  • Artificial Analysis · 32.8% direct comparison weight · independent of the model provider
Which benchmark families carry the weight?
  • ARC-AGI · 67.2% direct comparison weight
  • Terminal-Bench Science · 4.8% direct comparison weight
  • omniscience · 4.6% direct comparison weight
  • briefcase · 4.0% direct comparison weight
  • GDP.pdf · 4.0% direct comparison weight
  • Other families (6) · 15.5% 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.

What changes when a source is removed?
Removed sourceRank before → afterRank movementCohort remaining / lost supportScore after (change)
Artificial Analysis32 → 43Down 11230 remaining / 372 lost75.2 (-4.3 points)
ARC Prize35 → 16Up 19602 remaining / 0 lost87.9 (+8.3 points)
Datacurve35 → 36Down 1602 remaining / 0 lost78.0 (-1.6 points)
LiveCodeBench35 → 36Down 1602 remaining / 0 lost79.5 (-0.1 points)
Scale35 → 35No movement602 remaining / 0 lost80.3 (+0.7 points)
Terminal-Bench35 → 34Up 1602 remaining / 0 lost79.7 (+0.1 points)
Cognition35 → 37Down 2602 remaining / 0 lost78.3 (-1.3 points)
LiveBench35 → 36Down 1601 remaining / 1 lost80.3 (+0.7 points)
SWE-rebench35 → 36Down 1602 remaining / 0 lost79.4 (-0.2 points)
Epoch AI35 → 36Down 1588 remaining / 14 lost79.8 (+0.2 points)
Vals AI35 → 34Up 1601 remaining / 1 lost79.1 (-0.5 points)
Provider reports · Anthropic35 → 36Down 1602 remaining / 0 lost79.5 (-0.0 points)
Provider reports · Moonshot AI35 → 36Down 1602 remaining / 0 lost79.3 (-0.2 points)
Provider reports · Meta35 → 36Down 1602 remaining / 0 lost79.5 (-0.1 points)
Provider reports · Mistral35 → 35No movement602 remaining / 0 lost79.6 (+0.0 points)
Provider reports · OpenAI35 → 35No movement602 remaining / 0 lost79.6 (+0.0 points)
All provider reports35 → 37Down 2602 remaining / 0 lost79.3 (-0.3 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 evidenceRecomputed scoreChange
arc-agi87.6+8.0
enterprise-ops82.6+3.0
deepswe78.0-1.6
frontiercode78.3-1.3
arc-agi-378.3-1.3
terminal-bench-science78.5-1.1
mmmu-pro80.3+0.7
harvey79.0-0.6
critpt80.2+0.6
ifbench79.1-0.5
epoch-game-puzzles79.2-0.4
enigma-eval79.9+0.4
tau-bench79.2-0.3
simpleqa79.9+0.3
gpqa79.3-0.2
hle79.8+0.2
mmlu-pro79.8+0.2
terminal-bench79.4-0.2
livebench-instructions79.8+0.2
automationbench79.4-0.2
swe-atlas-qna79.4-0.2
swe-atlas-test-writing79.4-0.2
swe-rebench79.4-0.2
multi-swe-bench79.4-0.2
omniscience79.4-0.2
aa-lcr79.4-0.2
analyst-agent79.8+0.2
livecodebench79.5-0.1
livebench-language79.7+0.1
frontiermath79.7+0.1
vibe-code79.5-0.1
livebench-coding79.5-0.1
aime79.5-0.1
briefcase79.5-0.1
swe-atlas-refactoring79.5-0.1
vals-legal-research79.5-0.1
swe-bench-pro79.7+0.1
vals-tax-agent79.5-0.1
vals-code-migration79.5-0.1
livebench-reasoning79.5-0.1
gmmlu79.5-0.0
vals-excel-modeling79.6-0.0
livebench-data79.6+0.0
livebench-math79.6-0.0
math50079.6-0.0
gdp-pdf79.6-0.0
itbench79.6-0.0
vals-finance-agent79.6-0.0
apex-agents79.6-0.0
gdpval79.6+0.0
What evidence is missing from the fit?

Across all collected settings for this model: 55 observations, 55 contributing, 0 matched without graph weight and 0 excluded. Counts do not establish rank eligibility.

    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.

    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 / sourceLowMediumHighXHighMax
    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 2026-11-22.

    38.75%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    50%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    52.5%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    53.75%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    53.75%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    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 2026-11-22.

    1,229.43 Elo
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    1,494.88 Elo
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    1,540.77 Elo
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    1,580.04 Elo
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    1,612.07 Elo
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    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 2026-11-22.

    63.23%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    70.28%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    69.85%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    70.55%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    71.28%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    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 2026-11-22.

    19.87%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    21.4%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    23.37%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    26.32%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    28.92%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    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 2026-11-22.

    28.6%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    29.2%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    28.8%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    28.4%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    30%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    Harvey LAB-AA v1.1: Hallucination-Gated All-Pass RateArtificial Analysis · Stirrup; Harvey private dataset v1.1.0; 120 tasks across 24 practice areas; up to 200 turns; internet disabled; GPT-6 Sol, Grok 4.7 and Claude Opus 5.5 rubric judges; GPT-6 Sol High two-pass hallucination check; hallucination-gated all-pass averaged across judges
    Benchmark details

    Average 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.

    3.06%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    5%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    5.83%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    6.11%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    11.11%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    MLCR-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
    Benchmark details

    Share 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.

    32.22%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    31.67%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    33.89%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    35%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    37.22%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    AA-Omniscience AccuracyArtificial Analysis · 3000 questions; 1 attempt; correct-answer share; GPT-5.6 Luna medium grader
    Benchmark details

    AA-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.

    54.63%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    59.17%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    61.43%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    60.7%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    61.63%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    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 2026-11-22.

    37.14%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    60.48%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    67.62%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    66.67%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    67.62%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    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 2026-11-22.

    37.88%
    Reported settings

    GPT-6.2 Sol (Low); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-low

    56.06%
    Reported settings

    GPT-6.2 Sol (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-medium

    58.59%
    Reported settings

    GPT-6.2 Sol (High); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-high

    56.06%
    Reported settings

    GPT-6.2 Sol (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol-xhigh

    60.1%
    Reported settings

    GPT-6.2 Sol (Max); Model source: https://artificialanalysis.ai/models/gpt-6-2-sol

    ARC-AGI-2ARC Prize · v2 Semi-Private
    Benchmark details

    Original public board. Configuration and task coverage may differ; inspect source.

    Unit: percent. Source collected 2026-11-22.

    57.92%
    Reported settings

    GPT-6.2 Sol (Low); Source model ID: openai-gpt-6-2-sol-low; 120 tasks

    82.92%
    Reported settings

    GPT-6.2 Sol (Medium); Source model ID: openai-gpt-6-2-sol-medium; 120 tasks

    85.83%
    Reported settings

    GPT-6.2 Sol (High); Source model ID: openai-gpt-6-2-sol-high; 120 tasks

    85.83%
    Reported settings

    GPT-6.2 Sol (XHigh); Source model ID: openai-gpt-6-2-sol-xhigh; 120 tasks

    85.83%
    Reported settings

    GPT-6.2 Sol (Max); Source model ID: openai-gpt-6-2-sol-max; 120 tasks

    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