GPT-5.5 High
redUse when task complexity matters more than cost or stability.
Score: 75
Cost: $24.93
Runtime: 2141s
Tasks passed: 6/12
Best for: harder coding tasks
complex codingharder synthesiscross-file reasoning
model collection
GPT Models for Codex
Compare the current GPT model tiers inside Codex in one place instead of splitting the family across separate pages.
Intra-family comparison
What changes inside this model family?
Dimension
GPT-5.5 High
GPT-5.5 Medium
Primary tradeoff
more headroom
better balance
Stability
higher risk
safer default
Best for
hard coding
sustained execution
Models in this set
GPT Models for Codex
GPT-5.5 Medium
yellowUse as the default Codex lane when sustained value and steadier execution matter most.
Score: 87.5
Cost: $20.99
Runtime: 1950s
Tasks passed: 7/12
Best for: long-running work
long-running taskscost-aware engineeringiterative research
Platform
Comparison
Generated: 2026-07-01T12:30:00+08:00
Latest successful sync: 2026-07-01T12:20:00+08:00
Freshness: fresh_with_public_source_sync
FAQ
Direct answers for searchers.
Why combine these GPT models on one page?
Because the real buying question is usually high versus medium, not two isolated pages with duplicated context.
Which GPT model should be the default?
Current public signals make GPT-5.5 Medium the stronger default for sustained work, while High remains the special-case harder lane.