vibecodingnav
中/EN
Models/gpt-5-4

GPT 5.4

22nd on DeepSWE v1.1 Best (xhigh) at 51.8%.

OpenAI · Proprietary · Fetched 2026-10-10

DeepSWE v1.1
51.8%
#22 / 28 · 50.3–53.3

Benchmark scores

Boards older than 90 days are not counted as recommendations
BenchmarkSettingScoreRankCostAs of
DeepSWE v1.1xhigh · mini-swe-agent51.850.3–53.3#22$5.652026-09-22
SWE-bench Verified—This board has stopped updating——2026-02-26
Aider Polyglot—This board has stopped updating——2025-11-20

Use it or not

Why
  • Still has a checkable long-horizon engineering score on DeepSWE.
When not
  • Well behind the GPT 5.5 / 5.6 rows on that board.

Appears in these scenarios

Collected from scenario boards
Build a Web frontend#22DeepSWE 51.8%Build an Android app#22DeepSWE 51.8%Write CLI, scripts, and DevOps#22DeepSWE 51.8%

Peers

ModelDeepSWEPrice $/M
GPT 5.4Current51.8%—DeepSeek V4 Flash53.3%—Muse Spark 1.153.3%—Gemini 3.6 Flash46.7%—GLM 5.243.8%—