OpenAI · Released Mar 17, 2026GPT-5.4 nano
Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation.
- Capability
- 145.8ECI · #75 of 148
- Input
- $0.20per 1M tokens
- Output
- $1.25per 1M tokens
- Context
- 400K128K max output
What it costs
- Input
- $0.20per million tokens
- Output
- $1.25per million tokens
- Cached input
- $0.02per million tokens
- Blended (3:1)
- $0.463cheaper than 63% of priced models
Typical monthly bills
| Monthly usage | Estimated cost |
|---|---|
| Side project2M input + 0.5M output tokens | $1.02 |
| Team assistant25M input + 5M output tokens | $11.25 |
| Production app250M input + 50M output tokens | $112.50 |
Official OpenAI API price, as listed on models.dev.
Independent benchmarks
#75 of 148 scored models
The shaded band is Epoch AI’s confidence range (143.2–147.7); the tick marks the median scored model.
Scores from Epoch AI, run independently of OpenAI.
The details
- Lab
- OpenAI
- Released
- Mar 17, 2026
- Knowledge cutoff
- Aug 31, 2025
- Context window
- 400,000 tokens
- Max output
- 128,000 tokens
- Inputs
- Text, Images
- Output
- Text
- Reasoning
- Adjustable effort low · medium · high · xhigh
- Tool calling
- Yes
- Structured output
- Yes
- Weights
- Proprietary
- API model ID
gpt-5.4-nanoon OpenAI- Availability
- 26 API providerslisted on models.dev
What OpenAI claims
Published by the lab at launch. Settings vary, so compare these only with care.
| Benchmark | Score | Setting | Source |
|---|---|---|---|
| SWE-Bench Pro | 52.4 resolve rate | reasoning effort xhigh | Source |
| Terminal-Bench v2.0 | 46.3 | reasoning effort xhigh | Source |
| MCP Atlas | 56.1 | reasoning effort xhigh | Source |
| Toolathlon | 35.5 | reasoning effort xhigh | Source |
| τ²-Bench Telecom | 92.5 | reasoning effort xhigh | Source |
| GPQA Diamond | 82.8 | reasoning effort xhigh | Source |
| Humanity's Last Exam | 37.7 | with tools | Source |
| Humanity's Last Exam | 24.3 | without tools | Source |
| OSWorld-Verified | 39 success rate | reasoning effort xhigh | Source |
| MMMU Pro | 69.5 | with Python | Source |
| MMMU Pro | 66.1 | without tools | Source |
| OmniDocBench v1.5 | 0.2419 overall edit distance | reasoning effort none | Source |
More from OpenAI
About GPT-5.4 nano
How much does GPT-5.4 nano cost?
GPT-5.4 nano costs $0.20 input / $1.25 output per million tokens (official OpenAI API price). Cached input is $0.02 per million tokens. At a 3:1 input-to-output mix that is $0.463 per million tokens, cheaper than 63% of the 360 priced models we track.
What is the context window of GPT-5.4 nano?
GPT-5.4 nano accepts up to 400,000 tokens per request and can write up to 128,000 tokens in one response.
How good is GPT-5.4 nano?
Epoch AI gives GPT-5.4 nano a Capabilities Index score of 145.8 (likely range 143.2–147.7), ranking it #75 of 148 models Epoch has scored. Epoch AI benchmark results: GPQA Diamond 78.5%, FrontierMath Tiers 1–3 44.9%, OTIS Mock AIME 2024–2025 87.8%, SimpleQA Verified 11.7%.
Is GPT-5.4 nano open source?
No. GPT-5.4 nano is proprietary; you use it through OpenAI’s API or partner platforms.
What inputs does GPT-5.4 nano support?
GPT-5.4 nano accepts text and images and replies in text. It is a reasoning model with low, medium, high and xhigh effort settings, supports tool calling and can return structured JSON output.
When was GPT-5.4 nano released?
OpenAI released GPT-5.4 nano on Mar 17, 2026. Its training data runs to Aug 31, 2025.
What are the best alternatives to GPT-5.4 nano?
The closest current models from other labs on capability, price and release date are MiniMax-M2.7 (MiniMax, ECI 145.9, $0.30 / $1.20), Qwen3.6 35B-A3B (Alibaba (Qwen), ECI 143.9, $0.248 / $1.49), DeepSeek V3.2 (DeepSeek, ECI 146.3, $0.296 / $0.48) and Nemotron 3 Ultra 550B A55B (NVIDIA, ECI 146.2, $0.50 / $2.50).