This website uses cookies for authentication, support chat, and analytics. Only authentication cookies are turned on by default.
Animate this data
Bar chart race of xAI, Google DeepMind, OpenAI and 7 more, by GPQA Diamond.
Create it free
Takes a free account
xAI
xAI has risen from 53.8% in 2024 to 94% in 2026.
Aider Polyglot
80 %
Ale bench
1,508
Apex Agents
77 %
Arc Agi
90 %
Arc Agi 2
67 %
BALROG
44 %
Blueprint bench 2
33 %
Chess Puzzles
41 %
Cl bench
22 %
Cl bench Life
12 %
Critpt
20 %
CursorBench
46 %
DeepResearchBench
47 %
Deepswe
67 %
Dtbench
97 %
EBR bench
30 %
Forecastbench
66 %
FrontierCode
48 %
FrontierMath
20 %
FrontierMath Tier 4
2 %
FrontierMath Tier 4 V2
32 %
FrontierMath Tiers 1 3 V2
66 %
FrontierSWE
30 %
Furniture Assembly
40 %
GBAEval
65 %
GDPval
21 %
GPQA Diamond
94 %
Lech Mazur Writing
8 %
Live bench
54 %
Lmca
49 %
Math Level 5
91 %
MirrorCode
41 %
Mystery Game Puzzles
34 %
OTIS Mock AIME 2024 2025
99 %
PostTrainBench
23 %
Proofbench
51 %
SciCode
58 %
SimpleBench
76 %
SimpleQA Verified
56 %
Surface Evolver bench
74 %
TerminalBench
57 %
Vending bench 2
10,537
WebDev Arena
1,638
WeirdML
67 %
WeirdML V3
8 %
Training Compute
500,000,000,000,010.00T FLOP
xAI benchmark results over time.