GPT-5.4
OpenAIMULTIMODAL Proprietary
- Provider
- OpenAI
- Modality
- MULTIMODAL
- Parameters
- —
- Context window
- 1,050,000 tokens
- Weights
- Proprietary
- Released
- 5 Mar 2026
GPT-5.4 (March 2026) is the value-flagship: near-top quality on coding, research and math at roughly half the price of GPT-5.5, with native computer-use and a ~1M-token context.
Best for
- The default for most apps: complex coding, analysis, spreadsheets, hard math
- Agentic / computer-use workflows
- Professional knowledge work needing reliable accuracy
- Near-flagship quality on a budget
How it compares — and the India angle
- About 95% of GPT-5.5's everyday quality at half the cost — for most Indian apps this should be the default, not 5.5.
- Accepts text + image input but isn't an image generator — pair it with a dedicated image model.
- Step down to GPT-5.4 mini for cheap bulk traffic; up to GPT-5.5 only for the hardest agentic coding.
Benchmarks
Representative public scores (approximate, higher is better) for relative comparison. Check the provider for the latest official results.
MMLU90
GPQA Diamond93
SWE-bench Verified82
AIME (math)95
How to access
API ~$2.50 / 1M input, ~$15 / 1M output. Powers ChatGPT Plus (₹1,999/mo) 'Thinking' and Pro.
Access GPT-5.4