Gemini 4 Argon
Claude Opus 5.5
Muse Wonder
MiMo-V2.6-Pro
Qwen3.8-Max
Kimi K3
GLM-5.3
ERNIE 5.1
DeepSeek-V4.1-Flash-20260910
Grok 4.20 Beta
amazon-nova-experimental-chat
Dola Seed 2.0 Pro
NVIDIA-Nemotron-3-Ultra-550B-A55B
Gemma 4 31B
Step 5 Preview
GPT-5.4
GPT-5.4 is a proprietary model from OpenAI, released Mar 5, 2026. It ranks #33 of 351 on the LMArena text leaderboard with a rating of 1469. It costs $2.50 per million input tokens and $15 per million output tokens, with a 1.1M-token context window.
- Developer
- OpenAI
- Released
- Mar 5, 2026
- Context window
- 1.1M
- Price in / out, per 1M tokens
- $2.50 / $15
- Weights
- Proprietary
- Inputs
- Text, Images, Files
- Reasoning
- Yes
- Tool use
- Yes
The rating LMArena published at each update since Mar 11, 2026. Ratings move as votes come in and new models join.
Benchmarks
| Benchmark | Score bar | Accuracy | Rank | Cost / test |
|---|---|---|---|---|
| SWE-benchCoding | 78.20% | 29 / 85 | $0.80 | |
| Vibe Code BenchCoding | 48.47% | 61 / 115 | $7.51 | |
| LiveCodeBenchCoding | 84.14% | 41 / 143 | $2.5 / $15 | |
| MedCodeMedicine | 41.29% | 55 / 107 | $2.5 / $15 | |
| MedQAMedicine | 96.09% | 5 / 92 | $2.5 / $15 | |
| GPQA DiamondReasoning | 91.67% | 21 / 138 | $2.5 / $15 | |
| MMLU-ProReasoning | 87.48% | 27 / 138 | $2.5 / $15 | |
| AIMEReasoning | 96.67% | 5 / 96 | $2.5 / $15 | |
| MMMUReasoning | 87.51% | 15 / 93 | $2.5 / $15 |
Vals AI, as of Oct 24, 2026.
LMArena, by category
| Leaderboard | Rating | Rank | Votes |
|---|---|---|---|
| TextLMArena | 1469 | 33 / 351 | 64.8K |
| CodingLMArena | 1497 | 30 / 346 | 17.6K |
| MathLMArena | 1488 | 24 / 339 | 3.4K |
| Creative writingLMArena | 1439 | 45 / 349 | 11.1K |
| Instruction followingLMArena | 1469 | 30 / 351 | 21.9K |
| Hard promptsLMArena | 1485 | 28 / 351 | 42.1K |
| Multi-turnLMArena | 1482 | 21 / 349 | 12K |
| Long queriesLMArena | 1474 | 40 / 330 | 28.4K |
| ExpertLMArena | 1507 | 22 / 305 | 6.4K |
| Legal and governmentIndustries | 1486 | 24 / 325 | 5.2K |
| Medicine and healthcareIndustries | 1458 | 55 / 325 | 4.8K |
| Business and financeIndustries | 1475 | 18 / 344 | 13.1K |
| ChineseLanguages | 1519 | 33 / 328 | 4.1K |
| SpanishLanguages | 1452 | 57 / 249 | 2.1K |
| GermanLanguages | 1473 | 28 / 261 | 1.1K |
| FrenchLanguages | 1493 | 20 / 252 | 2.4K |
| JapaneseLanguages | 1485 | 16 / 236 | 676 |
| WebDevLMArena | 1464 | 74 / 124 | 1.4K |
| VisionLMArena | 1303 | 18 / 146 | 30.7K |
LMArena, CC BY 4.0, as of Oct 22, 2026. Where a model is rated at several effort levels, its best is shown (GPT-5.4 (High), GPT-5.4, GPT-5.4 (High, Codex Harness), GPT-5.4 (Medium, Codex Harness)).
Compare GPT-5.4
OpenAI
The lab behind GPT and ChatGPT, which brought large language models to hundreds of millions of people. Founded 2015, based in San Francisco. openai.com
Questions
Who makes GPT-5.4?
GPT-5.4 is made by OpenAI, founded 2015 and based in San Francisco.
How much does GPT-5.4 cost?
$2.50 per million input tokens and $15 per million output tokens, at its provider’s list price.
What is GPT-5.4’s context window?
1.1M tokens (1,050,000).
How does GPT-5.4 rank?
#33 of 351 on LMArena’s text leaderboard, rated 1469 (95% interval 1466–1473) from 64,757 votes, as of Oct 22, 2026.
Is GPT-5.4 open source?
No. GPT-5.4 is proprietary and served through its provider’s API.