OpenAI model product
GPT-5.4
4 is OpenAI's most capable and efficient frontier model for professional work.
Updated Aug 12, 2026. Default version: GPT-5.4
LLMBoard score77.1GPT-5.4
Coverage80%23 benchmark families
Context window1.1MTokens
Official input price$2.5OpenAI API
GPT-5.4 Specifications
Technical details for the model's default version.
- Version
- GPT-5.4
- Released
- Mar 5, 2026
- Knowledge cutoff
- Aug 31, 2025
- Parameters
- N/A
- Context window
- 1.1M
- Max output
- 128K
- Inputs
- image, pdf, text
- Outputs
- text
- Open weights
- No
- License
- Proprietary
GPT-5.4 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
GPT-5.4 Benchmark Results
Benchmark scores for GPT-5.4.
| Graphwalks parents <128k | 89.8% | 01 | 11 | 100.0% | C | |
| ARC-AGI | 93.7% | 02 | 7 | 83.3% | C | |
| Graphwalks BFS <128k | 93.0% | 02 | 11 | 90.0% | C | |
| ARC-AGI v2 | 73.3% | 03 | 16 | 86.7% | C | |
| LiveBench | 80.3% | 03 | 38 | 94.6% | B | |
| Tau2 Telecom | 98.9% | 03 | 35 | 94.1% | C | |
| FrontierMath | 47.6% | 04 | 17 | 81.3% | C | |
| Graphwalks parents >128k | 32.4% | 04 | 7 | 50.0% | C | |
| Finance Agent | 56.0% | 06 | 8 | 28.6% | C | |
| Terminal-Bench 2.0 | 75.1% | 06 | 49 | 89.6% | C | |
| OmniDocBench 1.5 | 89.1% | 07 | 17 | 62.5% | C | |
| Graphwalks BFS >128k | 21.4% | 08 | 11 | 30.0% | C | |
| Toolathlon | 54.6% | 08 | 31 | 76.7% | C | |
| FrontierSWE | 54.0% | 09 | 15 | 42.9% | B | |
| Legal Agent Benchmark | 0.4% | 09 | 13 | 33.3% | B | |
| MMMU-Pro | 81.2% | 09 | 66 | 87.7% | C | |
| GPQA | 92.8% | 11 | 234 | 95.7% | C | |
| OSWorld-Verified | 75.0% | 11 | 23 | 54.5% | C | |
| DeepSWE 1.1 | 52.0% | 14 | 20 | 31.6% | B | |
| BrowseComp | 82.7% | 19 | 58 | 68.4% | C | |
| SWE-Bench Pro | 57.7% | 20 | 45 | 56.8% | C | |
| MCP Atlas | 67.2% | 22 | 31 | 30.0% | C | |
| Humanity's Last Exam | 39.8% | 35 | 93 | 63.0% | C |
GPT-5.4 Arena Results
Preference and agent-evaluation results for the default version.
| search factuality | overall | 06 | 1212.1 | 87,310 | N/A | |
| text factuality | overall | 06 | 1485.3 | 60,739 | N/A | |
| agent tool hallucination | overall | 11 | 0.0 | N/A | 2.8M | |
| search style control | overall | 11 | 1194.6 | 87,364 | N/A | |
| agent steerability | overall | 12 | 0.1 | N/A | 58.3K | |
| search | overall | 14 | 1195.9 | 87,364 | N/A | |
| text factuality | overall | 14 | 1476.0 | 63,629 | N/A | |
| vision | overall | 14 | 1299.8 | 20,607 | N/A | |
| agent bash recovery steps | overall | 15 | 0.1 | N/A | 56.8K | |
| document | overall | 15 | 1470.1 | 29,809 | N/A | |
| agent | overall | 16 | 0.1 | N/A | 3M | |
| agent task outcome explicit | overall | 16 | 0.0 | N/A | 48.4K | |
| document style control | overall | 16 | 1469.3 | 29,809 | N/A | |
| agent praise complaint | overall | 18 | 0.0 | N/A | 14.6K | |
| vision style control | overall | 18 | 1282.6 | 20,607 | N/A | |
| vision | overall | 20 | 1293.0 | 20,516 | N/A | |
| vision style control | overall | 20 | 1280.5 | 20,516 | N/A | |
| text | overall | 21 | 1469.5 | 60,844 | N/A | |
| text style control | overall | 21 | 1476.5 | 60,844 | N/A | |
| text | overall | 34 | 1452.3 | 63,773 | N/A | |
| text style control | overall | 39 | 1465.5 | 63,773 | N/A | |
| webdev | overall | 39 | 1462.6 | 1,278 | N/A | |
| webdev | overall | 45 | 1442.2 | 1,318 | N/A | |
| webdev | overall | 64 | 1390.8 | 1,704 | N/A |
GPT-5.4 Runtime Performance
Provider-specific output speed and catalog latency for GPT-5.4. Runtime does not affect the capability score.
| OpenAI | 383.823 tok/s | 1.239 s | 1M | 128K |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
GPT-5.4 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- $2.5 input, $15 output per 1M
- Official provider
- OpenAI
- Lowest third-party
- From $0.75 input, $6 output per 1M via Xpersona
- Tracked offerings
- 39
| UnoRouter | gpt-5.4:free | global | N/A | N/A | 1.1M | |
| Xpersona | gpt-5.4 | global | $0.75 | $6 | 1.1M | |
| UnoRouter | gpt-5.4 | global | $1.8 | $10.8 | 1.1M | |
| Poe | openai/gpt-5.4 | global | $2.2 | $14 | 1.1M | |
| Impossibl | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| CrossModel | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Vivgrid | gpt-5.4 | global | $2.5 | $15 | 400K | |
| NanoGPT | openai/gpt-5.4 | global | $2.5 | $15 | 922K | |
| NEAR AI Cloud | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| DaoXE | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| LLM Gateway | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| OpenCode Zen | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| GitHub Copilot | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| 302.AI | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Neon | gpt-5-4 | global | $2.5 | $15 | 1.1M | |
| SAP AI Core | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Pioneer | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| AI-ROUTER | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Abacus | gpt-5.4 | global | $2.5 | $15 | 400K | |
| Azure Cognitive Services | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Azure | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| AIHubMix | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| FreeModel | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Kilo Gateway | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Merge Gateway | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Vercel AI Gateway | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| OpenRouter | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Cloudflare AI Gateway | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Perplexity Agent | openai/gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| OpenAI | gpt-5.4 | global | $2.5 | $15 | 1.1M | |
| Databricks | databricks-gpt-5-4 | global | $2.5 | $15 | 1.1M | |
| Amazon Bedrock | openai.gpt-5.4 | global | $2.75 | $16.5 | 272K | |
| Cortecs | gpt-5.4 | global | $2.9 | $15.45 | 1.1M | |
| Venice AI | openai-gpt-54 | global | $3.13 | $18.8 | 1M | |
| ZenMux | openai/gpt-5.4 | global | $3.75 | $18.75 | 1.1M | |
| OrcaRouter | openai/gpt-5.4 | global | $5 | $22.5 | 1.1M |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
GPT-5.4 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
| GPT-5.4 | 77.1 | N/A | 1.1M | 128K | No | Proprietary |
What is GPT-5.4?
Key information about GPT-5.4 and its available data.
4 is OpenAI's most capable and efficient frontier model for professional work.
It combines industry-leading coding capabilities with native computer-use, up to 1M tokens of context, full-resolution vision processing, tool search for large tool ecosystems, and improved reasoning across spreadsheets, presentations, and documents. It is the most token-efficient reasoning model in the GPT-5 series.
Data as of 2026-08-11.
GPT-5.4 vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to GPT-5.4
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
OP
GPT-5.6-Luna
OpenAI
OP
GPT-5.5-Pro
OpenAI
OP
GPT-5.2-Pro
OpenAI
OP
GPT-5.2
OpenAI
OP
GPT-5.5
OpenAI
OP
GPT-5.6-Terra
OpenAI
FAQ
Common questions about GPT-5.4.
When was GPT-5.4 released?
GPT-5.4's default version was released on Mar 5, 2026.
How much does GPT-5.4 cost?
GPT-5.4's official API price is $2.5 per million input tokens and $15 per million output tokens via OpenAI. The lowest tracked third-party offer starts at $0.75 input and $6 output via Xpersona.
Who created GPT-5.4?
GPT-5.4 was created by OpenAI.
What is the context window for GPT-5.4?
The default version has a 1.1M token context window.
Is GPT-5.4 open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer GPT-5.4?
39 provider offerings are linked to the default version.
What models should I compare GPT-5.4 with?
Nearby ranked alternatives include Seed 2.1 Pro, DeepSeek-V4-Flash, Gemini 3.1 Pro.