GPT-5.4 benchmark ranking on every AI leaderboard
closed weights · #45 overall, podium score 2.0 · top 10 on 2 of 12 · released Mar 4, 2026
12 of 12 boards
| Rank | Board | Score | |||
|---|---|---|---|---|---|
| 8 |  HLEReasoning · as “gpt-5.4-2026-03-05 (xhigh thinking)” | Reasoning | 36.2% | 42 | |
| 11 |  Document ArenaVision · as “gpt-5.4” | Vision | 1,471 | 38 | |
| 13 | Vision ArenaVision · as “gpt-5.4-high” | Vision | 1,285 | 100 | |
| 14 |  GPQA DiamondReasoning · as “gpt-5.4-2026-03-05_xhigh” | Reasoning | 93.3% | 100 | |
| 14 | Epoch ECIReasoning · as “GPT-5.4” | Reasoning | 156.9 | 100 | |
| 16 |  ARC-AGI-2Reasoning · as “GPT-5.4 (XHigh)” | Reasoning | 74.0% | 78 | |
| 16 |  LiveBenchReasoning · as “gpt-5.4-xhigh” | Reasoning | 78.0% | 59 | |
| 16 | Search ArenaAgents and tools · as “gpt-5.4-search” | Agents and tools | 1,197 | 33 | |
| 20 | MCP AtlasAgents and tools · as “gpt-5.4 (xhigh)” | Agents and tools | 70.6% | 34 | |
| 23 | Text ArenaChat · as “gpt-5.4-high” | Chat | 1,476 | 100 | |
| 47 | WebDev ArenaCoding · as “gpt-5.4-high (codex-harness)” | Coding | 1,463 | 100 | |
| 9 |  MMLU-ProReasoning · stale since 2026-03-11 · as “GPT-5.4” | Reasoning | 87.5% | 100 |
Category ranks and more models
Sources: each leaderboard's own published results page, captured every day at 04:30 UTC. Scores exactly as each board published them; we run no evaluations. Logos via logo.dev; trademarks belong to their owners.