OpenAI · Released Mar 5, 2026GPT-5.4
Agent-ready GPT for coding and computer-use workflows at a lower cost.
- Capability
- 156.9ECI · #16 of 148
- Input
- $2.50per 1M tokens
- Output
- $15.00per 1M tokens
- Context
- 1.05M128K max output
What it costs
- Input
- $2.50per million tokens
- Output
- $15.00per million tokens
- Cached input
- $0.25per million tokens
- Over 272K tokens
- $5.00 / $22.50input / output per million
- Blended (3:1)
- $5.63pricier than 90% of priced models
Typical monthly bills
| Monthly usage | Estimated cost |
|---|---|
| Side project2M input + 0.5M output tokens | $12.50 |
| Team assistant25M input + 5M output tokens | $137.50 |
| Production app250M input + 50M output tokens | $1,375 |
Official OpenAI API price, as listed on models.dev.
Independent benchmarks
#16 of 148 scored models
The shaded band is Epoch AI’s confidence range (154.9–159.1); the tick marks the median scored model.
Scores from Epoch AI, run independently of OpenAI.
The details
- Lab
- OpenAI
- Released
- Mar 5, 2026
- Knowledge cutoff
- Aug 31, 2025
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
- Inputs
- Text, Images, PDFs
- Output
- Text
- Reasoning
- Adjustable effort low · medium · high · xhigh
- Tool calling
- Yes
- Structured output
- Yes
- Weights
- Proprietary
- API model ID
gpt-5.4on OpenAI- Availability
- 36 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 | 59.1 resolve rate | — | Source |
| SWE-Atlas Codebase QnA | 40.8 | — | Source |
| SWE-Atlas Codebase QnA | 36.3 | — | Source |
| SWE-Atlas Refactoring | 44.29 | — | Source |
| SWE-Atlas Test Writing | 44.36 | — | Source |
| SWE-Atlas Test Writing | 40 | — | Source |
| Artificial Analysis Coding Agent Index | 53.6 | medium | Source |
| SWE-Atlas Codebase QnA | 72.4 | medium | Source |
| SWE-Bench Pro | 18.4 | medium | Source |
| Terminal-Bench v2.1 | 69.8 | medium | Source |
| Artificial Analysis Coding Agent Index | 52.2 | medium | Source |
| SWE-Atlas Codebase QnA | 72.9 | medium | Source |
More from OpenAI
About GPT-5.4
How much does GPT-5.4 cost?
GPT-5.4 costs $2.50 input / $15.00 output per million tokens (official OpenAI API price). Cached input is $0.25 per million tokens. Requests over 272K tokens are billed at $5.00 input / $22.50 output. At a 3:1 input-to-output mix that is $5.63 per million tokens, more expensive than 90% of the 360 priced models we track.
What is the context window of GPT-5.4?
GPT-5.4 accepts up to 1,050,000 tokens per request and can write up to 128,000 tokens in one response.
How good is GPT-5.4?
Epoch AI gives GPT-5.4 a Capabilities Index score of 156.9 (likely range 154.9–159.1), ranking it #16 of 148 models Epoch has scored. Epoch AI benchmark results: GPQA Diamond 93.3%, FrontierMath Tiers 1–3 78.6%, OTIS Mock AIME 2024–2025 97.8%, SWE-bench Verified 76.9%, SimpleQA Verified 45.1%.
Is GPT-5.4 open source?
No. GPT-5.4 is proprietary; you use it through OpenAI’s API or partner platforms.
What inputs does GPT-5.4 support?
GPT-5.4 accepts text, images and PDFs 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 released?
OpenAI released GPT-5.4 on Mar 5, 2026. Its training data runs to Aug 31, 2025.
What are the best alternatives to GPT-5.4?
The closest current models from other labs on capability, price and release date are Kimi K3 (Moonshot AI, ECI 157.6, $3.00 / $15.00), Gemini 3.1 Pro Preview (Google, ECI 154.8, $2.00 / $12.00), Claude Opus 4.7 (Anthropic, ECI 156.4, $5.00 / $25.00) and Qwen3.8 Max (Alibaba (Qwen), ECI 156.6, $2.00 / $6.00).