GPT-5.4 nano
Parameters
Undisclosed
Context Length
Undisclosed
Modality
Text
Architecture
Undisclosed
License
Proprietary
Release Date
17 Mar 2026
Knowledge Cutoff
-
API Pricing (per 1M)
Input: $0.20 · Output: $1.25
Evaluation Benchmarks
Rank
#70
| Benchmark | Score | Rank |
|---|---|---|
Mathematics | max 0.91 | 21 |
Reasoning | max 0.81 | 34 |
Agentic Coding | max 0.47 | 35 |
Data Analysis | max 0.68 | 38 |
General | max 0.70 | 41 |
LiveBench Average | max 0.70 | 41 |
Coding | max 0.71 | 42 |
Graduate-Level QA | 0.828 | 52 |
Coding Index | max 0.56 | 72 |
Agentic Index | max 0.18 | 81 |
General Text | high 1402 | 94 |
Intelligence Index | max 0.21 medium 0.20 Standard 0.12 | 116 121 172 |
Rankings
Overall Rank
#70
Coding Rank
#92
About GPT-5.4 nano
GPT-5.4 nano is OpenAI's smallest and cheapest GPT-5.4 variant, optimized for tasks where speed and cost matter most. A significant upgrade over GPT-5 nano, recommended for classification, data extraction, ranking, and coding subagents handling simpler supporting tasks. Pricing: $0.20/M input, $1.25/M output. Available via API as gpt-5.4-nano. Released March 17, 2026.
Technical Specifications
Architecture specifications are undisclosed for proprietary models.
Attention
Attention Structure
Multi-Head Attention
Attention Heads
Undisclosed
Key-Value Heads
Undisclosed
Attention Head Dimension
Undisclosed
Position Embedding
Absolute Position Embedding
RoPE Theta
Undisclosed
Sliding Window Attention
Undisclosed
Sliding Window Size
Undisclosed
Sliding Window Ratio
Undisclosed
Linear Attention
Undisclosed
Linear Attention Ratio
Undisclosed
Normalization
Undisclosed
Activation Function
Undisclosed
Dimensions
Hidden Dimension Size
Undisclosed
Number of Layers
Undisclosed
FFN Intermediate Size (Dense)
Undisclosed
Multi-Token Prediction Heads
Undisclosed
Tokenizer
Vocabulary Size
Undisclosed
Resources
About GPT-5.4
GPT-5.4 is OpenAI's most capable and efficient frontier model for professional work, combining the industry-leading coding capabilities of GPT-5.3-Codex with major advances in reasoning, computer use, and agentic workflows. It introduces native computer-use capabilities, tool search for large tool ecosystems, substantially improved knowledge work (spreadsheets, presentations, documents), and is OpenAI's most factual and token-efficient reasoning model. Supports up to 1M context tokens in Codex. Released March 5, 2026.