GPT-5.4
GPT-5.4 is a closed AI model from OpenAI with an index score of 166 (GPT-4 = 100), rank 35 of 217.
166150–182Index score#35 of 217 modelsPro mode: 171 · $30 / $180
$2.5/ $15Price per million tokensinput / output · via OpenRouter
1.05MContext windowtokens it reads at once
5 Mar 2026ReleasedReplaced by GPT-5.5
Where it sitson the frontier
Among the first models to reach this score.
Test resultsby area · tick = best by any model
Reasoning & knowledgeevidence
Mathsevidence
Codingevidence
Agentsevidence
Professional workevidence
Novel problemsevidence
Long tasksevidence
Third-party results only, at the model’s best standard setting. Hover a bar for who ran it.
What others sayrank on 12 leaderboards
The Model Indexthird-party tests, our method#35of 217 · 166
LMArena · Mathsmaths questions in chat#24of 346 · 1491LMArena · Codingcoding questions in chat#30of 357 · 1521LMArena · Textoverall chat quality, judged by people#32of 362 · 1475LMArena · Instruction followingfollowing detailed instructions#34of 362 · 1471LMArena · Long promptslong, complex requests#39of 340 · 1481Epoch Capabilities Indexoverall capability from many benchmarks#28of 231 · 157.9LMArena · Visionunderstanding images#19of 145 · 1282LMArena · Creative writingcreative writing#50of 360 · 1440LMArena · Documentsworking with long documents#12of 40 · 1473LMArena · Searchanswering with web search#21of 41 · 1198LMArena · WebDevbuilding web apps and front ends#71of 133 · 1462Best attop three among current models
Where to use itprices checked 18 Nov 2026
API providers 3 · per 1M tokens, in / out
Host prices from OpenRouter; going direct can differ.
Facts and sources
Released5 Mar 2026
Announcementnot linked yet
Release entry–
Scored asGPT-5.4
Pro modeindex 171 · $30 / $180 per M
Output speed–
Hugging Faceclosed weights