RankingEffort benchmarks

GPT-6 Astra · XHigh

Compare models

180 measured results · 7 effort settings · 83 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

AGPT-6 Astra · XHigh3 of 7 capabilities supported

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

Agentic
A85.8Supported
Hard reasoning
A83.6Supported
Coding
A93.1Supported
Human pref
Unknown
Knowledge
A70.0Preliminary
Multimodal
A73.9Preliminary
Long context
A53.6Preliminary

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 Astra · XHigh: 95.9 overall comparison score. Meets the overall evidence requirements.

6 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?

OpenAI · 25.0% direct comparison weight · provider report
Cognition · 18.4% direct comparison weight · independent of the model provider
ARC Prize · 18.0% direct comparison weight · independent of the model provider
Artificial Analysis · 16.3% direct comparison weight · independent of the model provider
Datacurve · 11.1% direct comparison weight · independent of the model provider
Scale · 8.5% direct comparison weight · independent of the model provider
Terminal-Bench · 1.5% direct comparison weight · independent of the model provider
Epoch AI · 1.0% direct comparison weight · independent of the model provider

Which benchmark families carry the weight?

FrontierCode · 18.4% direct comparison weight
DeepSWE · 16.1% direct comparison weight
ARC-AGI · 9.3% direct comparison weight
arc-agi-3 · 8.7% direct comparison weight
AutomationBench · 6.3% direct comparison weight
Other families (16) · 41.1% 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.

DeepSWE v1.1 review: Epoch AI reported grading errors in at least 23 of 113 tasks. These published results remain included. Review their limitations in Epoch's review when interpreting this evidence.

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 2026-10-20.

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	1 → 1	No movement	203 remaining / 241 lost	96.9 (+1.0 points)
ARC Prize	1 → 3	Down 2	444 remaining / 0 lost	96.1 (+0.1 points)
Datacurve	1 → 1	No movement	444 remaining / 0 lost	95.4 (-0.5 points)
LiveCodeBench	1 → 1	No movement	444 remaining / 0 lost	95.9 (-0.0 points)
Scale	1 → 1	No movement	444 remaining / 0 lost	96.1 (+0.1 points)
Terminal-Bench	1 → 1	No movement	444 remaining / 0 lost	95.9 (-0.0 points)
Cognition	1 → 1	No movement	444 remaining / 0 lost	94.6 (-1.3 points)
LiveBench	1 → 1	No movement	443 remaining / 1 lost	96.4 (+0.4 points)
SWE-rebench	1 → 1	No movement	444 remaining / 0 lost	95.9 (-0.1 points)
Epoch AI	1 → 1	No movement	431 remaining / 13 lost	95.9 (-0.1 points)
Vals AI	1 → 1	No movement	443 remaining / 1 lost	95.9 (-0.1 points)
Provider reports · Anthropic	1 → 1	No movement	444 remaining / 0 lost	95.9 (-0.0 points)
Provider reports · Moonshot AI	1 → 1	No movement	444 remaining / 0 lost	96.0 (+0.0 points)
Provider reports · Meta	1 → 1	No movement	444 remaining / 0 lost	95.9 (-0.0 points)
Provider reports · Mistral	1 → 1	No movement	444 remaining / 0 lost	95.9 (+0.0 points)
Provider reports · OpenAI	1 → 1	No movement	444 remaining / 0 lost	93.1 (-2.8 points)
All provider reports	1 → 1	No movement	444 remaining / 0 lost	93.3 (-2.6 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
frontiercode	94.6	-1.3
arc-agi	96.8	+0.9
deepswe	95.4	-0.5
arc-agi-3	95.6	-0.3
gdpval	96.1	+0.1
aa-lcr	96.0	+0.1
gpqa	95.8	-0.1
livebench-instructions	96.0	+0.1
simpleqa	96.0	+0.1
mmmu-pro	96.0	+0.1
ifbench	95.9	-0.1
swe-rebench	95.9	-0.1
epoch-game-puzzles	95.9	-0.1
swe-atlas-test-writing	95.9	-0.1
terminal-bench	95.9	-0.1
briefcase	96.0	+0.1
automationbench	95.9	-0.0
frontiermath	95.9	-0.0
mmlu-pro	96.0	+0.0
gdp-pdf	95.9	-0.0
livebench-data	95.9	-0.0
hle	96.0	+0.0
swe-atlas-qna	95.9	-0.0
livebench-reasoning	95.9	-0.0
enigma-eval	96.0	+0.0
vibe-code	95.9	-0.0
omniscience	96.0	+0.0
vals-code-migration	95.9	-0.0
livebench-language	95.9	-0.0
swe-bench-pro	95.9	-0.0
tau-bench	95.9	+0.0
vals-excel-modeling	95.9	-0.0
vals-finance-agent	95.9	+0.0
terminal-bench-science	95.9	-0.0
livebench-coding	95.9	+0.0
harvey	95.9	-0.0
critpt	95.9	-0.0
aime	95.9	-0.0
apex-agents	95.9	+0.0
livecodebench	95.9	-0.0
gmmlu	95.9	-0.0
math500	95.9	-0.0
multi-swe-bench	95.9	-0.0
enterprise-ops	95.9	+0.0
analyst-agent	95.9	-0.0
vals-legal-research	95.9	-0.0
itbench	95.9	-0.0
swe-atlas-refactoring	95.9	+0.0
livebench-math	95.9	+0.0
vals-tax-agent	95.9	-0.0

What evidence is missing from the fit?

Across all collected settings for this model: 311 observations, 174 contributing, 0 matched without graph weight and 137 excluded. Counts do not establish rank eligibility.

11 · Benchmark family or source protocol has not been reviewed
5 · 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
1 · Source-reported label. OpenAI documents Low through Max for Astra; this does not establish a supported non-reasoning API setting.
2 · Source-reported None does not establish a supported Astra API setting.
1 · Starred protocol annotation requires row-specific review before ranking.
8 · Evaluator has not been reviewed for independent admission
2 · Effort is not identified
1 · The task table records 169 trials of the nominal 170.
18 · Provider-published claim. Independent source results are admitted separately with exact checkpoint, effort and compatible protocol; this grid does not establish a matched comparison.
1 · Provider-published claim. Independent source results are admitted separately with exact checkpoint, effort and compatible protocol; this grid does not establish a matched comparison. Historical provider claim; secondary and superseded for current FrontierSWE comparisons by the independent Proximal FrontierSWE v2 mean@5 board at https://www.frontierswe.com/. Version-unspecified claims are not v2 results. Excluded from weighted Overall.
24 · Supporting evidence outside the reviewed capability core
14 · No matched, identified effort unit; duplicates and unnamed comparators do not add evidence
7 · Composite index or publisher-specific internal evaluation lacks a stable general-capability comparison unit.
25 · Narrow specialist biological, medical or security evaluation; preserved as raw evidence outside general capability scoring.
3 · No matched opponent in this evaluation unit
5 · Private user-flagged prompt protocol needs separate review before capability scoring; retained as an exact lower-is-better sourced claim.
4 · Safety failure behavior is not a capability or intelligence score.
1 · Secondary Artificial Analysis leaderboard report; retain the original evaluator once, without treating the launch reprint as additional evidence.
4 · 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	Low	Medium	High	XHigh	Max	None	Unspecified
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 2026-10-18.	Not reported	Not reported	Not reported	Not reported	51.25% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
AA-Briefcase EloArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Briefcase v1.1 is an agentic knowledge work benchmark developed by Artificial Analysis. AA-Briefcase Elo is a combined metric that aggregates rubric pass rate, analytical quality Elo and presentation Elo · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-10-18.	1,257.01 Elo Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	1,460.92 Elo Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	1,505.91 Elo Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	1,546.7 Elo Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	1,568.99 Elo Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
AA-LCR v1.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	80% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	79.67% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	80% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	80% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	80.67% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
AutomationBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of task objectives completed with no guardrail violations · Higher is better · Benchmark developed by Zapier · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	59.1% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	64.59% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	66.61% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	67.18% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	68.49% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
CritPtArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by Argonne and UIUC, with contributions from 60+ researchers globally. CritPt is under review by its authors following external feedback · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	26.29% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	29.14% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	28.86% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	31.43% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	31.71% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
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 2026-10-18.	30.4% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	30.4% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	31% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	32.2% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	31% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
GDPval-AA v2.1 LeaderboardArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Elo rating for performance on real-world work tasks · Anchored to DeepSeek V4.1 Flash (max) at 1600 · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-10-18.	1,389.25 Elo Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	1,486.4 Elo Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	1,522.92 Elo Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	1,555.91 Elo Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	1,572.84 Elo Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
GPQA DiamondArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	93.13% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	93.94% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	94.95% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	96.26% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	96.06% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
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 2026-10-18.	Not reported	Not reported	Not reported	Not reported	8.61% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
Humanity's Last ExamArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	49.21% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	52.73% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	53.06% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	54.59% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	54.68% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
ITBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Combined public and private Site Reliability Engineering (SRE) tasks · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	48.59% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
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 2026-10-18.	Not reported	Not reported	Not reported	Not reported	35% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
MMMU-ProArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	84.62% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	85.09% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	86.42% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	86.24% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	86.88% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
AA-Omniscience IndexArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. · Evaluation results measured independently by Artificial Analysis Unit: index. Source collected 2026-10-18.	40.55 Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	42.22 Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	43.73 Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	43.42 Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	43.4 Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
AA-Omniscience AccuracyArtificial Analysis · 6000 questions; 1 attempt; correct-answer share; GPT-5.6 Luna medium grader Benchmark details AA-Omniscience Accuracy (higher is better) measures the proportion of correctly answered questions out of all questions, regardless of whether the model chooses to answer · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	59.53% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	60.57% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	61.13% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	61.85% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	62.6% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
SciCodeArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark curated by scientists across 16 disciplines. SciCode is under review following an independent audit of the dataset · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	54.05% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	54.17% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	55.44% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	55.67% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	56.48% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
𝜏³-BankingArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by Sierra Research · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	31.96% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	35.46% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	40% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	43.09% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	41.44% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
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 2026-10-18.	Not reported	Not reported	Not reported	Not reported	63.33% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
Terminal-Bench v2.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by the Laude Institute and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	88.01% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	89.51% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	89.89% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	89.14% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	88.39% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	Not reported
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 2026-10-18.	41.92% Reported settings GPT-6 Astra (Low); Model source: https://artificialanalysis.ai/models/gpt-6-astra-low	49.49% Reported settings GPT-6 Astra (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-astra-medium	54.04% Reported settings GPT-6 Astra (High); Model source: https://artificialanalysis.ai/models/gpt-6-astra-high	59.6% Reported settings GPT-6 Astra (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-astra-xhigh	59.09% Reported settings GPT-6 Astra (Max); Model source: https://artificialanalysis.ai/models/gpt-6-astra	Not reported	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 2026-10-18.	85.42% Reported settings GPT-6 Astra (Low); Source model ID: openai-gpt-6-astra-low; 120 tasks	92.08% Reported settings GPT-6 Astra (Medium); Source model ID: openai-gpt-6-astra-medium; 120 tasks	92.08% Reported settings GPT-6 Astra (High); Source model ID: openai-gpt-6-astra-high; 120 tasks	93.33% Reported settings GPT-6 Astra (XHigh); Source model ID: openai-gpt-6-astra-xhigh; 120 tasks	95% Reported settings GPT-6 Astra (Max); Source model ID: openai-gpt-6-astra-max; 120 tasks	59.58% Reported settings GPT-6 Astra (None); Source model ID: openai-gpt-6-astra-none; 120 tasks Source-reported label. OpenAI documents Low through Max for Astra; this does not establish a supported non-reasoning API setting. Source label needs review	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 2026-10-18.	67% Reported settings gpt-6-astra; Reported effort: low; 113 unique tasks; 452 attempts; 4 runs	72.8% Reported settings gpt-6-astra; Reported effort: medium; 113 unique tasks; 452 attempts; 4 runs	73.2% Reported settings gpt-6-astra; Reported effort: high; 113 unique tasks; 452 attempts; 4 runs	74.1% Reported settings gpt-6-astra; Reported effort: xhigh; 113 unique tasks; 452 attempts; 4 runs	73.2% Reported settings gpt-6-astra; Reported effort: max; 113 unique tasks; 452 attempts; 4 runs	Not reported	Not reported
Terminal-Bench 2.1Terminal-Bench · Codex Benchmark details Original public board. Configuration and task coverage may differ; inspect source. Unit: percent. Source collected 2026-10-18.	86.74% Reported settings GPT-6 Astra; Reported effort: low; 445 trials; recorded 2026-09-03	86.97% Reported settings GPT-6 Astra; Reported effort: medium; 445 trials; recorded 2026-09-03	87.42% Reported settings GPT-6 Astra; Reported effort: high; 445 trials; recorded 2026-09-03	85.84% Reported settings GPT-6 Astra; Reported effort: xhigh; 445 trials; recorded 2026-09-03	86.74% Reported settings GPT-6 Astra; Reported effort: max; 445 trials; recorded 2026-09-03	Not reported	Not reported
ARC-AGI-3 · provider-adapterARC Prize · Provider Adapter Benchmark details ARC-AGI-3; Provider Adapter. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-10-18.	98.03% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-astra; variant provider-adapter-low; public-demo task scores are excluded	98.44% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-astra; variant provider-adapter-medium; public-demo task scores are excluded	99.95% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-astra; variant provider-adapter-high; public-demo task scores are excluded	98.44% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-astra; variant provider-adapter-xhigh; public-demo task scores are excluded	98.55% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-astra; variant provider-adapter-max; public-demo task scores are excluded	96.72% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-astra; variant provider-adapter-none; public-demo task scores are excluded Source-reported None does not establish a supported Astra API setting. Source label needs review	Not reported
ARC-AGI-3 · standardARC Prize · Standard Benchmark details ARC-AGI-3; Standard. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-10-18.	17.45% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-astra; variant low; public-demo task scores are excluded	38.59% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-astra; variant medium; public-demo task scores are excluded	54.82% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-astra; variant high; public-demo task scores are excluded	59.34% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-astra; variant xhigh; public-demo task scores are excluded	62.71% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-astra; variant max; public-demo task scores are excluded	35.18% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-astra; variant none; public-demo task scores are excluded Source-reported None does not establish a supported Astra API setting. Source label needs review	Not reported

What these comparisons mean

Each source reports its own effort labels. Matching labels do not establish equal compute budgets or identical fallback behavior. Open Reported settings to inspect the exact model name, agent and task coverage. Multiple results at one level remain visible rather than selecting the best. Source-reported None or Non-reasoning labels are not treated as evidence that an API supports those settings.

These measurements use their own dated source collection. The leaderboard fits identified configurations jointly; it contains no pooled model ratings. Explicit thinking budgets are separate configurations. Read the methodology.

Download all effort benchmark results · 8,915 results across 259 models · Collected 2026-10-18

GPT-6 Astra · XHigh

Compare with

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context

Capability score evidence

Benchmark results