OpenAIUnited States· released Sep 2026
Best published result on 4 of the 24 tests it has taken.
OpenAI, United States. One of 30 models from this lab on the Index.
Snapshot November 10, 2026
OpenCharts Index
88.2
Ranked #4 of 80 · 11.8% behind the frontier on average · 24 comparable tests · 5 counting categories
Drop any single test and the Index lands between 86.9 and 89.2.
Needs more results (3 of 5 so far)
Averaging 99.2 on the 3 it has; the score opens at 4 of 5.
Distance to today's frontier on the hardest tests. Not a measure of intelligence.
What it's known for
Strongest in math (99.4) and reasoning (93.1). It holds the best published result on ARC-AGI-1, FrontierMath Tier 4 and GDP.pdf and 1 more. It trails the frontier most in agentic, 26.4% behind on average. Between 10% and 15% behind the best published results, on average.
Strongest category
Math
Competition and research mathematics, graded on the final answer or the proof.
99.4
Mean of 4 of the category's 4 comparable tests. Counts toward the Index.
Best single result
Solve the original Abstraction and Reasoning Corpus: small grids, a handful of examples, one hidden rule.
98.5%
Best published result of 67 models.
The first ARC set is close to saturated at the frontier, so it now shows whether smaller models can generalize at all.
Where it trails
Play a little-known card-based adventure game ten times through a text interface, keeping notes, and score on the last two games.
54.3%
40.0% behind the best published result (90.5%) · #7 of 30.
Getting better at an unfamiliar job by doing it is what people do in their first weeks at work. This checks whether practice helps a model at all.
Not measured yet
Covered, but not averaged into the Index:
See it at work
Every test it has taken, by category, with the kind of task it faced, who produced the number and where the result landed against the best published one. The examples are original and illustrative, never items from the datasets themselves.
Math · mean 99.4 across 4 comparable tests · counts toward the Index
Solve competition mathematics problems that have a single exact integer answer.
100.0%
best published result
of 98 models
max
45 problems: one problem is 2.2 points, so gaps of a few points are within noise.
Source: Epoch AI (internal runs)Solve the hardest FrontierMath tier: problems that take expert mathematicians days.
100.0%
best published result
of 68 models
max
48 problems: one problem is about 2 points. Same OpenAI commissioning disclosure as tiers 1 to 3.
Source: Epoch AI (internal runs)Write a complete, rigorous proof, graded for rigor rather than for a final answer.
99.0%
99.0% of the best (100.0%)
#5 of 81
Solve unpublished research-level mathematics problems written by professional mathematicians (tiers 1 to 3).
93.7%
98.5% of the best (95.1%)
#2 of 78 · n = 290
max
Commissioned by OpenAI, which has access to much of the problem set; Epoch discloses this and keeps a holdout it does not share.
Source: Epoch AI (internal runs)Every result is read against the best published one on its test: percent scores as a share of the best above chance, Arena ratings as the expected win rate against the board leader, open-ended values on a log scale. Sitting a harder exam never lowers a model. Every number links to the publisher that produced it.
Head to head
Choose a rival. Every counted test both have taken appears side by side as a share of the best published result on that test, with the category means above. A win is a higher share; a gap under 2 points is a tie, because one item on a 45-problem exam is 2.2 points and publishers' own standard errors are of that order.
GPT-6.1 Sol
#4·Index 88.2A-
4 : 10
wins · 10 ties · 24 shared
GPT-6 Astra
#1·Index 93.7A
GPT-6 Astra comes out ahead on 10 of the 24 tests both have taken.
Category means
Reasoning
Knowledge
Coding
Math
Agentic
Multimodal
Human preference
Shared tests · biggest gaps first
Bars are distances to the best published result on each test (percent scores above chance, Arena ratings as win rate against the leader, open-ended values on a log scale). Raw values as the publishers report them.
Open GPT-6 AstraScores
The best published run per benchmark, the raw value as the publisher reports it, who produced the number, how close it comes to the best published result on that test, and where the model sits among every model scored on it. Percent scores are read above chance, Arena ratings as a win rate against the board leader, open-ended values on a log scale. A test with too few models to compare against is shown but not counted, and so is an archived test: one whose publisher no longer covers new models.
Category means
Mean · tests taken of the category's comparable tests. A faded category either rests on fewer than half of them or is a single-test category (marked “beside”), which is shown next to the Index and never averaged into it.
95.4%
98.6% of the best (96.4%), above chance
96th percentile of 96 · n = 198
max
Source: Epoch AI — AI Benchmarking Hub94.2%
99.1% of the best (95.0%)
98th percentile of 67 · n = 120
max
Source: Epoch AI — AI Benchmarking Hub98.5%
best published result
of 67 models
xhigh
Source: Epoch AI — AI Benchmarking Hub31.7%
88.1% of the best (36.0%)
96th percentile of 102
max
Source: Epoch AI — AI Benchmarking Hub68.0%
85.1% of the best (79.0%), above chance
96th percentile of 76 · n = 100
max
Source: Epoch AI — AI Benchmarking Hub80.0%
87.6% of the best (90.0%), above chance
97th percentile of 72 · n = 100
max
Source: Epoch AI — AI Benchmarking Hub73.9%
78.6% of the best (94.0%)
96th percentile of 75 · n = 1,000
max
Source: Epoch AI — AI Benchmarking Hub55.8%
83.4% of the best (66.9%)
71th percentile of 96 · n = 338
high
Source: Epoch AI — AI Benchmarking Hub50.6%
93.6% of the best (54.0%)
96th percentile of 25 · n = 11
xhigh · agent: codex_cli · reasoning effort: xhigh
Source: Epoch AI — AI Benchmarking Hub50.2%
91.9% of the best (54.6%)
85th percentile of 41
medium · harness: codex · reasoning effort: medium
Source: Epoch AI — AI Benchmarking Hub1,743
39% expected win rate against the board leader (1,818)
95th percentile of 110
max · board 2026-11-09
Source: Arena (LMArena) — Leaderboard Dataset100.0%
best published result
of 98 models
max
Source: Epoch AI — AI Benchmarking Hub93.7%
98.5% of the best (95.1%)
97th percentile of 78 · n = 290
max
Source: Epoch AI — AI Benchmarking Hub100.0%
best published result
of 68 models
max
Source: Epoch AI — AI Benchmarking Hub99.0%
99.0% of the best (100.0%)
91th percentile of 81
60.0%
71.3% of the best (84.2%)
72th percentile of 58
max
Source: Epoch AI — AI Benchmarking Hub$12,861
92.9% of the best ($16,484) on a log scale
94th percentile of 65
54.3%
60.0% of the best (90.5%)
79th percentile of 30 · n = 21
max
Source: Epoch AI — AI Benchmarking Hub23.8%
70.1% of the best (34.0%)
88th percentile of 9
max
Source: Epoch AI — AI Benchmarking Hub1,293
48% expected win rate against the board leader (1,309)
87th percentile of 71
max · board 2026-11-09
Source: Arena (LMArena) — Leaderboard Dataset80.0%
83.3% of the best (90.0%), above chance
92th percentile of 37 · n = 60
max
Source: Epoch AI — AI Benchmarking Hub43.2%
73.7% of the best (58.6%)
89th percentile of 36 · n = 50
35.0%
best published result
of 37 models
max
Source: Epoch AI — AI Benchmarking Hub1,482
44% expected win rate against the board leader (1,526)
79th percentile of 113
max · board 2026-11-09
Source: Arena (LMArena) — Leaderboard DatasetNearby on the leaderboard
29 of the models on this page run inside Theo today. Theo picks the right one for each step and always shows which engine answered.