GPT-5.5-Cyber

Unverified OpenAI language model release

Requested OpenAI language model release name could not be verified in accessible first-party sources; recorded as an unconfirmed target.

Pricing

Model profile · Current snapshot

Input / 1M

$5.00

Output / 1M

$30.00

Blended / 1M

$11.25

Output TPS

63.68

TTFT (s)

25.22

Model Intelligence

Arena ranking

Benchmarkable

Yes

Model level

release

Intelligence Index

58.9

Coding Index

58.5

GPQA

0.93

HLE

0.43

SciCode

0.56

IFBench

0.72

LCR

0.73

TerminalBench Hard

0.6

TAU2

0.93

Recent stories

1 linked story

releasePRIMARY2026-05-07

OpenAI rolls out GPT-5.5-Cyber limited preview for critical-infrastructure defenders

OpenAI introduced GPT-5.5-Cyber in limited preview for defensive security teams and paired it with GPT-5.5 plus Trusted Access for Cyber. The release matters because OpenAI is separating cyber-specific access and permissiveness from general-model access rather than treating security work as a normal prompting mode.