RankingEffort benchmarks

GPT-5.4 · XHigh effort benchmarks

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

Model

GPT-5.4

74 measured results · 5 reported effort labels · 51 benchmark/harness combinations. Model evidence and sources →

Capability score evidence

GPT-5.4 · XHigh: 38.3 overall comparison score. Meets the overall evidence requirements.

Which sources carry the weight?

Datacurve · 15.7% direct comparison weight · independent
Artificial Analysis · 15.2% direct comparison weight · independent
Scale · 15.2% direct comparison weight · independent
Moonshot AI · 15.0% direct comparison weight · provider report
ARC Prize · 14.6% direct comparison weight · independent
LiveBench · 14.3% direct comparison weight · independent
Vals AI · 10.0% 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	41.2	+2.9
ARC Prize	38.0	-0.3
Datacurve	37.1	-1.2
LiveCodeBench	38.3	-0.0
Scale	40.4	+2.1
Terminal-Bench	38.4	+0.1
Cognition	39.4	+1.1
LiveBench	36.6	-1.7
SWE-rebench	37.8	-0.5
Epoch AI	38.6	+0.2
Vals AI	39.2	+0.9
Provider reports · Anthropic	38.3	+0.0
Provider reports · Moonshot AI	38.6	+0.3
Provider reports · Meta	38.3	-0.0
Provider reports · Mistral	38.3	+0.0
All provider reports	38.6	+0.3

Every benchmark family

Removed evidence	Recomputed score	Change
terminal-bench	42.0	+3.7
mmmu-pro	40.2	+1.9
deepswe	37.1	-1.2
frontiercode	39.4	+1.1
harvey	39.1	+0.8
swe-rebench	37.8	-0.5
arc-agi	37.8	-0.5
simpleqa	38.7	+0.4
epoch-game-puzzles	37.9	-0.4
livebench-data	37.9	-0.4
swe-atlas-test-writing	38.0	-0.3
gpqa	38.0	-0.3
livebench-instructions	38.6	+0.2
vals-finance-agent	38.1	-0.2
livebench-math	38.5	+0.2
swe-atlas-qna	38.5	+0.2
livebench-coding	38.5	+0.2
apex-agents	38.1	-0.2
vals-code-migration	38.1	-0.2
multi-swe-bench	38.1	-0.2
vals-legal-research	38.1	-0.2
vibe-code	38.1	-0.2
hle	38.5	+0.2
aa-lcr	38.2	-0.2
briefcase	38.5	+0.1
vals-excel-modeling	38.2	-0.1
enigma-eval	38.4	+0.1
mmlu-pro	38.4	+0.1
itbench	38.4	+0.1
gdp-pdf	38.4	+0.1
gdpval	38.4	+0.1
swe-bench-pro	38.2	-0.1
frontiermath	38.4	+0.1
automationbench	38.4	+0.1
ifbench	38.2	-0.1
aime	38.2	-0.1
arc-agi-3	38.2	-0.1
swe-atlas-refactoring	38.4	+0.1
tau-bench	38.2	-0.1
livebench-reasoning	38.3	-0.1
scicode	38.4	+0.1
enterprise-ops	38.4	+0.1
critpt	38.3	-0.0
livecodebench	38.3	-0.0
omniscience	38.3	+0.0
math500	38.3	-0.0
livebench-language	38.3	-0.0
analyst-agent	38.3	+0.0
gmmlu	38.3	-0.0

What evidence is missing from the fit?

Across all collected settings for this model: 110 observations, 73 contributing, 0 matched without graph weight and 37 excluded. Counts do not establish rank eligibility.

3 · Benchmark family or source protocol has not been reviewed
11 · Cited comparator requires original evaluation provenance; not a new matched run.
15 · Supporting evidence outside the reviewed capability core
7 · No reviewed effort for this catalog observation; any separately collected effort measurement is counted separately
1 · Microsoft report Table 11 and section 4.1: MAI Terminal-Bench removes timeouts and uses a minimal ReAct harness; comparator values are cited from other model releases. A common evaluation protocol is not established.

Inspect every observation and exclusion →

Capability profile

One configuration, seven capabilities

Comparison score · 0–100

AGPT-5.4 · XHigh5 of 7 capabilities supported

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

Agentic
A52.1Supported
Hard reasoning
A54.3Supported
Coding
A33.6Supported
Human pref
Unknown
Knowledge
A40.1Supported
Multimodal
A42.6Supported
Long context
A63.7Preliminary

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	Low	Medium	High	XHigh	Non-reasoning
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 2026-09-12.	Not reported	Not reported	Not reported	33.26% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	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 2026-09-12.	76.67% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	82% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	58.33% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-non-reasoning
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.	7.43% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	23.43% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	0.57% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-non-reasoning
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	1,306.71 Elo Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	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 2026-09-12.	87.07% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	92.02% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	74.85% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-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 2026-09-12.	30.82% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	43.74% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	11.26% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-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 2026-09-12.	65.92% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	73.95% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	48.44% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-non-reasoning
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.	77.98% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	78.44% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	70.58% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-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 2026-09-12.	4.78 Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	5.78 Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	-15.67 Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-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 2026-09-12.	47.85% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	50.85% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	37.38% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-non-reasoning
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.	74.56% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	87.13% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	35.96% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-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 2026-09-12.	Not reported	Not reported	Not reported	39.59% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	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 2026-09-12.	43.18% Reported settings GPT-5.4 (low); Model source: https://artificialanalysis.ai/models/gpt-5-4-low	Not reported	Not reported	57.58% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	37.88% Reported settings GPT-5.4 (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-5-4-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 2026-09-12.	Not reported	Not reported	Not reported	78.28% Reported settings GPT-5.4 (xhigh); Model source: https://artificialanalysis.ai/models/gpt-5-4	Not reported
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.	29.17% Reported settings GPT-5.4 (Low); Source model ID: gpt-5-4-low; 120 tasks	55.42% Reported settings GPT-5.4 (Medium); Source model ID: gpt-5-4-medium; 120 tasks	67.5% Reported settings GPT-5.4 (High); Source model ID: gpt-5-4-high; 120 tasks	73.95% Reported settings GPT-5.4 (XHigh); Source model ID: gpt-5-4-xhigh; 120 tasks	Not reported
DeepSWE v1.1Datacurve · mini-swe-agent Benchmark details Original public board. Configuration and task coverage may differ; inspect source. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	51.8% Reported settings gpt-5-4; Reported effort: xhigh; 113 unique tasks; 452 attempts; 4 runs	Not reported
SWE-bench Pro publicScale · mini-swe-agent starred rows Benchmark details Original public board. Configuration and task coverage may differ; inspect source. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	59.1% Reported settings gpt-5.4 (xHigh)*; 731 tasks; source date 2026-04-08	Not reported
SWE Atlas · QnA · codexScale · codex Benchmark details 124 code comprehension tasks; strict all-rubrics pass metric; Opus 4.5 judge; source-reported agent harness. Repositories drawn from SWE-bench Pro; distinct tasks. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	40.8% Reported settings 124 code comprehension tasks; strict all-rubrics pass metric; Opus 4.5 judge; source-reported agent harness. Repositories drawn from SWE-bench Pro; distinct tasks.; published row 2026-03-09T16:27:54.000Z; reported margin ±5.1 percentage points; No contamination annotation reported (not proof of absence)	Not reported
SWE Atlas · Refactoring · codexScale · codex Benchmark details 70 refactoring tasks across 10 repositories; behavior preservation and task-specific grading; source-reported agent harness. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	44.29% Reported settings 70 refactoring tasks across 10 repositories; behavior preservation and task-specific grading; source-reported agent harness.; published row 2026-05-06T01:44:16.000Z; reported margin ±6.76 percentage points; No contamination annotation reported (not proof of absence)	Not reported
SWE Atlas · Test Writing · codexScale · codex Benchmark details 90 test-writing tasks; execution, mutation checks and rubric judging; source-reported agent harness. Repositories drawn from SWE-bench Pro; distinct tasks. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	44.36% Reported settings 90 test-writing tasks; execution, mutation checks and rubric judging; source-reported agent harness. Repositories drawn from SWE-bench Pro; distinct tasks.; published row 2026-03-26T18:24:53.000Z; reported margin ±6.04 percentage points; No contamination annotation reported (not proof of absence)	Not reported
SWE Atlas · Test Writing · mini-sweScale · mini-swe Benchmark details 90 test-writing tasks; execution, mutation checks and rubric judging; source-reported agent harness. Repositories drawn from SWE-bench Pro; distinct tasks. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	40% Reported settings 90 test-writing tasks; execution, mutation checks and rubric judging; source-reported agent harness. Repositories drawn from SWE-bench Pro; distinct tasks.; published row 2026-03-26T18:24:57.000Z; reported margin ±6 percentage points; No contamination annotation reported (not proof of absence)	Not reported
LiveBench · AMPS Hard · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	98% Reported settings LiveBench official task evaluation; livebench-2026-06-25-AMPS_Hard; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · code completion · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	80.44% Reported settings LiveBench official task evaluation; livebench-2026-06-25-code_completion; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · code generation · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	74.65% Reported settings LiveBench official task evaluation; livebench-2026-06-25-code_generation; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · connections · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	100% Reported settings LiveBench official task evaluation; livebench-2026-06-25-connections; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · consecutive events · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	86.19% Reported settings LiveBench official task evaluation; livebench-2026-06-25-consecutive_events; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · integrals with game · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	93% Reported settings LiveBench official task evaluation; livebench-2026-06-25-integrals_with_game; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · javascript · 2026-06-25LiveBench · LiveBench Mini-SWE-Agent, 250-step limit Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	68.18% Reported settings LiveBench Mini-SWE-Agent, 250-step limit; livebench-2026-06-25-javascript; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · logic with navigation · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	68% Reported settings LiveBench official task evaluation; livebench-2026-06-25-logic_with_navigation; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · math comp · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	94.12% Reported settings LiveBench official task evaluation; livebench-2026-06-25-math_comp; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · olympiad · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	91.47% Reported settings LiveBench official task evaluation; livebench-2026-06-25-olympiad; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · paraphrase · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	63.43% Reported settings LiveBench official task evaluation; livebench-2026-06-25-paraphrase; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · plot unscrambling · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	65.9% Reported settings LiveBench official task evaluation; livebench-2026-06-25-plot_unscrambling; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · python · 2026-06-25LiveBench · LiveBench Mini-SWE-Agent, 250-step limit Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	50% Reported settings LiveBench Mini-SWE-Agent, 250-step limit; livebench-2026-06-25-python; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · simplify · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	70% Reported settings LiveBench official task evaluation; livebench-2026-06-25-simplify; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · spatial · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	98% Reported settings LiveBench official task evaluation; livebench-2026-06-25-spatial; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · story generation · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	79.02% Reported settings LiveBench official task evaluation; livebench-2026-06-25-story_generation; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · summarize · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	68.42% Reported settings LiveBench official task evaluation; livebench-2026-06-25-summarize; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · tablejoin · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	51.75% Reported settings LiveBench official task evaluation; livebench-2026-06-25-tablejoin; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · tablereformat · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	100% Reported settings LiveBench official task evaluation; livebench-2026-06-25-tablereformat; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · theory of mind · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	88.46% Reported settings LiveBench official task evaluation; livebench-2026-06-25-theory_of_mind; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · typescript · 2026-06-25LiveBench · LiveBench Mini-SWE-Agent, 250-step limit Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	43.33% Reported settings LiveBench Mini-SWE-Agent, 250-step limit; livebench-2026-06-25-typescript; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · typos · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	82% Reported settings LiveBench official task evaluation; livebench-2026-06-25-typos; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
LiveBench · zebra puzzle · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	98% Reported settings LiveBench official task evaluation; livebench-2026-06-25-zebra_puzzle; gpt-5.4-xhigh; checkpoint=gpt-5.4	Not reported
SWE-rebench · 2026-02 · 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	62.81% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2026-02-tools; gpt-5.4-2026-03-05-medium; checkpoint=gpt-5.4	Not reported	Not reported	Not reported
SWE-rebench · 2026-03 · 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	57.83% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2026-03-tools; gpt-5.4-2026-03-05-medium; checkpoint=gpt-5.4	Not reported	Not reported	Not reported
SWE-rebench · 2026-04 · 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	49.77% Reported settings SWE-rebench standardized ReAct tools; swe-rebench-2026-04-tools; gpt-5.4-2026-03-05-medium; checkpoint=gpt-5.4	Not reported	Not reported	Not reported
Vals AI · Code Migration v1Vals AI · Vals Code Migration shared agent harness Benchmark details Hidden behavior-test pass rate, equally weighted across 40 repositories (30 CLI and 10 COBOL-to-Java). Anti-cheat failures score zero. Code-quality ratings and language splits are not additional votes. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	34.98% Reported settings {"checkpoint":"gpt-5.4-2026-03-05","evaluatorModel":"openai/gpt-5.4-2026-03-05","sourceUpdatedOn":"2026-09-10","reasoning_effort":"xhigh","compute_effort":null,"reasoning":null,"temperature":null,"top_p":null,"max_output_tokens":128000,"verbosity":null,"provider":"OpenAI","harness":null}	Not reported
Vals AI · MMLU-Pro v1Vals AI · Vals MMLU-Pro five-shot chain-of-thought; regex answer grading Benchmark details Mean accuracy across 14 subjects under Vals five-shot prompting. Subject splits share one family and are not additional votes. Vals labels this benchmark archived; it is historical corroboration, not ongoing frontier evaluation. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	87.48% Reported settings {"checkpoint":"gpt-5.4-2026-03-05","evaluatorModel":"openai/gpt-5.4-2026-03-05","sourceUpdatedOn":"2026-09-01","reasoning_effort":"xhigh","compute_effort":null,"reasoning":null,"temperature":null,"top_p":null,"max_output_tokens":128000,"verbosity":null,"provider":"OpenAI","harness":null}	Not reported
Vals AI · Vibe Code Bench v1.1Vals AI · Vals modified OpenHands harness; five hours or 1,000 turns per app Benchmark details Mean app test pass rate on the canonical v1.1 test set. Browser Use executes workflow tests; each test requires at least 90% of its substeps. Automated judging can introduce error. Alternate agents are excluded. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	67.42% Reported settings {"checkpoint":"gpt-5.4-2026-03-05","evaluatorModel":"openai/gpt-5.4-2026-03-05","sourceUpdatedOn":"2026-09-10","reasoning_effort":"xhigh","compute_effort":null,"reasoning":null,"temperature":null,"top_p":null,"max_output_tokens":128000,"verbosity":null,"provider":"OpenAI","harness":"OpenHands"}	Not reported
Vals AI · Harvey Legal Agent Benchmark v1Vals AI · Vals Harvey held-out legal file-work protocol; internet disabled Benchmark details Mean task resolution across two judges (GPT-5.5 Medium and Sonnet 4.6), requiring every criterion to pass. Criteria pass rates and practice-area splits do not add votes. Shared Harvey family; Fable 5 fallback run excluded. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	0% Reported settings {"checkpoint":"gpt-5.4-2026-03-05","evaluatorModel":"openai/gpt-5.4-2026-03-05","sourceUpdatedOn":"2026-09-10","reasoning_effort":"xhigh","compute_effort":null,"reasoning":null,"temperature":null,"top_p":null,"max_output_tokens":128000,"verbosity":null,"provider":"OpenAI","harness":null}	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 · 7,424 results across 240 models · Collected 2026-09-12

GPT-5.4 · XHigh effort benchmarks

GPT-5.4

Capability score evidence

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context