BREAKING
Grok 4.5 and GPT-5.6 Ship Same Week
0
T
V9 parameters
0
x
prior size
0
t/s
speed
Grok 4.5 Coding Benchmarks
Terminal Bench
83.3
Coding Agent
76
SWE Bench Pro
64.7
Two Frontier Lineups
Grok 4.5
xAI
●
Trained on Cursor workflows
●
Opus-class at lower cost
●
500k token context
GPT-5.6
OpenAI
●
Sol, Terra and Luna tiers
●
Coding, science, security focus
●
Up to 750 tokens/sec on Cerebras
0
x
fewer tokens vs Opus
0
%
τ³-Banking
0
$
per M input
Race Shifts to Real-World Workflows
AI NEWS BLITZ
xAI and OpenAI released new flagship models within a day of each other.