RankingEffort benchmarks

GPT-5 Mini effort benchmarks

Compare measured Low, Medium, High and other settings. Every number keeps its source.

Model

GPT-5 Mini

60 measured results · 4 reported effort labels · 29 benchmark/harness combinations. Model evidence and sources →

Capability score evidence

GPT-5 Mini · High: 10.9 overall comparison score. Meets the overall evidence requirements.

Which sources carry the weight?

Artificial Analysis · 40.2% direct comparison weight · independent
ARC Prize · 31.7% direct comparison weight · independent
SWE-rebench · 28.2% direct comparison weight · independent

These shares describe the comparisons entering the fit, not fractions of the final score.

How sensitive is this score?

Remove one evaluator or family and refit the graph against the same overall panel. Values use the matching dataset audited 2026-09-12. These are sensitivity checks, not confidence intervals; ranking eligibility may also change.

Removed evidence	Recomputed score	Change
Artificial Analysis	7.9	-3.0
ARC Prize	13.2	+2.3
Datacurve	9.3	-1.6
LiveCodeBench	10.9	-0.0
Scale	12.0	+1.0
Terminal-Bench	10.9	-0.0
Cognition	11.5	+0.6
LiveBench	11.0	+0.1
SWE-rebench	11.4	+0.5
Epoch AI	11.1	+0.2
Vals AI	10.5	-0.4
Provider reports · Anthropic	10.9	+0.0
Provider reports · Moonshot AI	10.9	-0.0
Provider reports · Meta	10.9	-0.0
Provider reports · Mistral	10.9	+0.0
All provider reports	10.9	-0.0

Every benchmark family

Removed evidence	Recomputed score	Change
arc-agi	13.1	+2.2
deepswe	9.3	-1.6
mmmu-pro	11.6	+0.7
frontiercode	11.5	+0.6
swe-rebench	11.4	+0.5
ifbench	10.5	-0.4
hle	11.3	+0.4
critpt	11.3	+0.4
simpleqa	11.3	+0.4
mmlu-pro	11.2	+0.3
omniscience	10.6	-0.3
scicode	11.2	+0.3
epoch-game-puzzles	10.7	-0.2
terminal-bench	11.1	+0.2
aa-lcr	10.7	-0.2
gpqa	10.7	-0.2
livebench-instructions	11.1	+0.2
arc-agi-3	10.8	-0.2
briefcase	11.1	+0.2
multi-swe-bench	10.8	-0.2
aime	10.8	-0.1
livecodebench	10.8	-0.1
livebench-language	11.0	+0.1
gmmlu	10.8	-0.1
swe-atlas-qna	10.8	-0.1
apex-agents	11.0	+0.1
harvey	10.8	-0.1
frontiermath	11.0	+0.1
livebench-coding	10.8	-0.1
vibe-code	10.8	-0.1
livebench-reasoning	10.8	-0.1
vals-finance-agent	10.9	-0.0
swe-atlas-test-writing	10.9	-0.0
tau-bench	10.9	-0.0
gdpval	10.9	+0.0
vals-code-migration	10.9	-0.0
vals-legal-research	10.9	-0.0
analyst-agent	10.9	+0.0
enterprise-ops	10.9	+0.0
vals-excel-modeling	10.9	-0.0
math500	10.9	+0.0
livebench-data	10.9	+0.0
swe-atlas-refactoring	10.9	-0.0
livebench-math	10.9	+0.0
gdp-pdf	10.9	-0.0
itbench	10.9	+0.0
enigma-eval	10.9	+0.0
swe-bench-pro	10.9	+0.0
automationbench	10.9	+0.0

What evidence is missing from the fit?

Across all collected settings for this model: 78 observations, 57 contributing, 0 matched without graph weight and 21 excluded. Counts do not establish rank eligibility.

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

Inspect every observation and exclusion →

Capability profile

One configuration, seven capabilities

Comparison score · 0–100

AGPT-5 Mini · High4 of 7 capabilities supported

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

Agentic
A15.6Supported
Hard reasoning
A11.9Supported
Coding
A10.9Supported
Human pref
Unknown
Knowledge
A22.6Supported
Multimodal
A13.0Preliminary
Long context
A28.0Preliminary

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.

Compare this configuration →Explore the model evidence profile →

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	Low	Medium	High
AA-Briefcase EloArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Briefcase 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 2026-09-12.	Not reported	Not reported	Not reported	437.58 Elo Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	46.67% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	85% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	90.67% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	39.33% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	72.33% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	72.33% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	Not reported	Not reported	Not reported	6.49% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	0% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	1.43% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	0% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
GDP.pdf: All-passArtificial 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 2026-09-12.	Not reported	Not reported	Not reported	8.4% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
GDPval-AA v2 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 a human baseline of 1,000 · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-09-12.	Not reported	Not reported	Not reported	880.27 Elo Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
Global-MMLU-LiteArtificial 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 2026-09-12.	Not reported	Not reported	Not reported	87.43% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	68.69% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	80.3% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	82.83% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	5.05% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	15.94% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	21.46% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	45.58% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	71.16% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	75.44% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	54.5% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	69.21% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	83.81% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	77.45% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	82.78% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	83.73% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	58.38% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	68.84% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	70.06% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	-53.05 Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	-11.03 Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	-17.32 Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	19.17% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	23.07% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	24.97% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
SciCodeArtificial 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 2026-09-12.	Not reported	Not reported	Not reported	39% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	31.87% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	71.05% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	68.42% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	Not reported	Not reported	Not reported	15.46% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	14.39% Reported settings GPT-5 mini (minimal); Model source: https://artificialanalysis.ai/models/gpt-5-mini-minimal	Not reported	28.79% Reported settings GPT-5 mini (medium); Model source: https://artificialanalysis.ai/models/gpt-5-mini-medium	33.33% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	Not reported	Not reported	Not reported	3.75% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
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 2026-09-12.	Not reported	Not reported	Not reported	0% Reported settings GPT-5 mini (high); Model source: https://artificialanalysis.ai/models/gpt-5-mini
ARC-AGI-2ARC Prize · v2 Semi-Private Benchmark details Original public board. Configuration and task coverage may differ; inspect source. Unit: percent. Source collected 2026-09-12.	1.67% Reported settings GPT-5 Mini (Minimal); Source model ID: gpt-5-mini-2025-08-07-minimal; 120 tasks	0.83% Reported settings GPT-5 Mini (Low); Source model ID: gpt-5-mini-2025-08-07-low; 120 tasks	4.03% Reported settings GPT-5 Mini (Medium); Source model ID: gpt-5-mini-2025-08-07-medium; 120 tasks	4.44% Reported settings GPT-5 Mini (High); Source model ID: gpt-5-mini-2025-08-07-high; 120 tasks
SWE-rebench · 2025-08 · toolsSWE-rebench · SWE-rebench standardized ReAct tools Benchmark details Average resolved rate across repeated runs, not best-of-N success. Disjoint complete calendar months within each model’s tested range; standardized harness and all-language task set. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	43.08% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2025-08-tools; gpt-5-mini-2025-08-07-medium; checkpoint=gpt-5-mini	Not reported
SWE-rebench · 2025-09 · toolsSWE-rebench · SWE-rebench standardized ReAct tools Benchmark details Average resolved rate across repeated runs, not best-of-N success. Disjoint complete calendar months within each model’s tested range; standardized harness and all-language task set. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	37.14% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2025-09-tools; gpt-5-mini-2025-08-07-medium; checkpoint=gpt-5-mini	Not reported
SWE-rebench · 2025-10 · toolsSWE-rebench · SWE-rebench standardized ReAct tools Benchmark details Average resolved rate across repeated runs, not best-of-N success. Disjoint complete calendar months within each model’s tested range; standardized harness and all-language task set. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	32.55% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2025-10-tools; gpt-5-mini-2025-08-07-medium; checkpoint=gpt-5-mini	Not reported
SWE-rebench · 2025-11 · toolsSWE-rebench · SWE-rebench standardized ReAct tools Benchmark details Average resolved rate across repeated runs, not best-of-N success. Disjoint complete calendar months within each model’s tested range; standardized harness and all-language task set. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	44.26% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2025-11-tools; gpt-5-mini-2025-08-07-medium; checkpoint=gpt-5-mini	Not reported
SWE-rebench · 2025-12 · toolsSWE-rebench · SWE-rebench standardized ReAct tools Benchmark details Average resolved rate across repeated runs, not best-of-N success. Disjoint complete calendar months within each model’s tested range; standardized harness and all-language task set. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	40.66% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2025-12-tools; gpt-5-mini-2025-08-07-medium; checkpoint=gpt-5-mini	44.31% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2025-12-tools; gpt-5-mini-2025-08-07-high; checkpoint=gpt-5-mini
SWE-rebench · 2026-01 · toolsSWE-rebench · SWE-rebench standardized ReAct tools Benchmark details Average resolved rate across repeated runs, not best-of-N success. Disjoint complete calendar months within each model’s tested range; standardized harness and all-language task set. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	30.83% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2026-01-tools; gpt-5-mini-2025-08-07-medium; checkpoint=gpt-5-mini	35% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2026-01-tools; gpt-5-mini-2025-08-07-high; checkpoint=gpt-5-mini

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 · 7,424 results across 240 models · Collected 2026-09-12

GPT-5 Mini effort benchmarks

GPT-5 Mini

Capability score evidence

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context