Unlock exclusive introductory pricing for the newly launched Gemini 3.5 Flash.
O

GPT-5.4

Input:$2/M
Output:$12/M
Cache Read:$0.2/M
Context:1,050,000
Max Output:128,000
Released:Mar 5, 2026

GPT-5.4 is the frontier model for complex professional work. Reasoning.effort supports: none (default), low, medium, high and xhigh.

New
Popular
Commercial Use

Playground for GPT-5.4

Explore GPT-5.4's Playground โ€” an interactive environment to test models, run queries in real time. Try prompts, adjust parameters, and iterate instantly to accelerate development and validate use cases.

Technical Specifications of GPT-5.4-2026-03-05

ItemGPT-5.4-2026-03-05
Model familyGPT-5
ProviderOpenAI
Release dateMarch 5, 2026
Context window1,050,000 tokens
Max output tokens128,000
Input typesText, Image
Output typesText
AudioNot supported
Reasoning controlsnone, low, medium, high, xhigh
Tool supportWeb search, File search, Code interpreter, Image generation
Knowledge cutoffAug 31, 2025
Snapshot stabilityLocked model behavior

What is GPT-5.4?

GPT-5.4 is a unifying frontier release that merges improvements from recent reasoning and coding lines (including the GPT-5.3-Codex work) into a single model targeted at professional knowledge work. It is positioned as a โ€œThinkingโ€ model for deeper, steerable reasoning and a โ€œProโ€ variant for the highest performance/throughput customers. Key themes of the release are: (1) longer context and document-scale understanding, (2) improved tool and โ€œcomputer useโ€ capabilities (controlling apps, spreadsheet/presentation editing), and (3) reduced factual errors and stronger multi-step planning.

Main features of GPT-5.4

  • Huge long-context capability (1M+ tokens experimental): GPT-5.4 supports experimental 1.05M token sessions (with pricing/limits) enabling whole-book / whole-codebase reasoning and multi-document synthesis. For general availability the standard window remains โ‰ˆ272K tokens.
  • Improved multi-step tool use & native โ€œcomputer useโ€: better desktop/browser control for agentic workflows (keyboard/mouse via a computer-use interface), web search that persists across rounds, and a new Tool Search mechanism to find connectors/tools efficiently. OpenAI reports state-of-the-art success on multiple computer-use and web-agent benchmarks.
  • Spreadsheet, document, and presentation generation/editing: specific tuning for office workflows; internal benchmarks show major gains on spreadsheet modelling and presentation quality. OpenAI also launched a ChatGPT for Excel add-in alongside the release.
  • Steerability & reasoning modes: โ€œThinkingโ€ mode produces an explicit plan/preamble for long tasks and supports mid-response steering (adjusting instructions during generation). Reasoning effort levels let users trade latency for deeper chain-of-thought reasoning.
  • Enhanced multimodal understanding: better interpretation of high-resolution images and charts (image input), used for document understanding and presentations.
  • Safety posture: OpenAI treats GPT-5.4 as a high-cyber-capability model and deploys enhanced safeguards similar to the GPT-5.3-Codex mitigations.

Benchmark performance

GPT-5.4ย GPT-5.3-CodexGPT-5.2
GDPval (wins or ties)83.0%70.9%70.9%
SWE-Bench Pro (Public)57.7%56.8%55.6%
OSWorld-Verified75.0%74.0%*ย 47.3%
Toolathlon54.6%51.9%46.3%
BrowseComp82.7%77.3%65.8%

GPT-5.4 vs Comparable Models

ModelContext WindowKey Strength
GPT-5.4-2026-03-051,050,000 tokensFrontier reasoning + agent workflows
GPT-5.3 InstantSmallerFaster everyday tasks
Claude Opus / Sonnet~200k tokensLong-form reasoning
Gemini 3 Pro~1M tokensMultimodal reasoning

Key difference: GPT-5.4 focuses heavily onย professional productivity workflows and agent capabilities, particularly when integrated with external tools.

Representative production use cases

  1. Enterprise document & compliance workflows: processing long contracts, extracting obligations, and drafting commentaries across multi-document corpora (benefits from the 272Kโ†’1M context options for single-session synthesis).
  2. Spreadsheet automation & financial modelling: generating formulas, building multi-sheet models from plain-English spec, reconciling inputs โ€” OpenAI reports large gains on junior investment-banking style tasks.
  3. Agentic automation & โ€œcomputer useโ€: automated browser / desktop workflows (installation, QA, tool orchestration) and multi-step tool chains (Zapier integrations cited as a use partner).
  4. Software engineering & code maintenance: code generation, refactorings, and terminal/CLI agent tasks (Terminal-Bench gains reported). For large codebases, the long context window helps but must be validated on task heuristics.
  5. Knowledge worker augmentation: research synthesis (BrowseComp improvements), slide generation and visual design for presentations.

How to access GPT-5.4 API

Step 1: Sign Up for API Key

Log in toย cometapi.com. If you are not our user yet, please register first. Sign into yourย CometAPI console. Get the access credential API key of the interface. Click โ€œAdd Tokenโ€ at the API token in the personal center, get the token key: sk-xxxxx and submit.

cometapi-key

Step 2: Send Requests to GPT-5.4 API

Select the โ€œgpt-5.4โ€ endpoint to send the API request and set the request body. The request method and request body are obtained from our website API doc. Our website also provides Apifox test for your convenience. Replace <YOUR_API_KEY> with your actual CometAPI key from your account. base url isย Chat Completions and Responses.

Insert your question or request into the content fieldโ€”this is what the model will respond to . Process the API response to get the generated answer.

Step 3: Retrieve and Verify Results

Process the API response to get the generated answer. After processing, the API responds with the task status and output data.

FAQ

Pricing for GPT-5.4

Explore competitive pricing for GPT-5.4, designed to fit various budgets and usage needs. Our flexible plans ensure you only pay for what you use, making it easy to scale as your requirements grow. Discover how GPT-5.4 can enhance your projects while keeping costs manageable.

ModelTierConditionComet Price (USD / M Tokens)Official Price (USD / M Tokens)Discount
gpt-5.4short_contextlen <= 272000
Input:$2.0000/M
Output:$12.0000/M
Cache Read:$0.2000/M
Input:$2.5000/M
Output:$15.0000/M
Cache Read:$0.2500/M
-20%
long_context-
Input:$4.0000/M
Output:$18.0000/M
Cache Read:$0.4000/M
Input:$5.0000/M
Output:$22.5000/M
Cache Read:$0.5000/M
-20%
gpt-5.4-2026-03-05short_contextlen <= 272000
Input:$2.0000/M
Output:$12.0000/M
Cache Read:$0.2000/M
Input:$2.5000/M
Output:$15.0000/M
Cache Read:$0.2500/M
-20%
long_context-
Input:$4.0000/M
Output:$18.0000/M
Cache Read:$0.4000/M
Input:$5.0000/M
Output:$22.5000/M
Cache Read:$0.5000/M
-20%
gpt-5.4-ministandard-
Input:$0.6000/M
Output:$3.6000/M
Cache Read:$0.0600/M
Input:$0.7500/M
Output:$4.5000/M
Cache Read:$0.0750/M
-20%
gpt-5.4-mini-2026-03-17standard-
Input:$0.6000/M
Output:$3.6000/M
Cache Read:$0.0600/M
Input:$0.7500/M
Output:$4.5000/M
Cache Read:$0.0750/M
-20%
gpt-5.4-nanostandard-
Input:$0.1600/M
Output:$1.0000/M
Cache Read:$0.0160/M
Input:$0.2000/M
Output:$1.2500/M
Cache Read:$0.0200/M
-20%
gpt-5.4-nano-2026-03-17standard-
Input:$0.1600/M
Output:$1.0000/M
Cache Read:$0.0160/M
Input:$0.2000/M
Output:$1.2500/M
Cache Read:$0.0200/M
-20%
gpt-5.4-proshort_contextlen <= 272000
Input:$24.0000/M
Output:$144.0000/M
Input:$30.0000/M
Output:$180.0000/M
-20%
long_context-
Input:$48.0000/M
Output:$216.0000/M
Input:$60.0000/M
Output:$270.0000/M
-20%
gpt-5.4-pro-2026-03-05short_contextlen <= 272000
Input:$24.0000/M
Output:$144.0000/M
Input:$30.0000/M
Output:$180.0000/M
-20%
long_context-
Input:$48.0000/M
Output:$216.0000/M
Input:$60.0000/M
Output:$270.0000/M
-20%

Sample code and API for GPT-5.4

Access comprehensive sample code and API resources for GPT-5.4 to streamline your integration process. Our detailed documentation provides step-by-step guidance, helping you leverage the full potential of GPT-5.4 in your projects.

curl https://api.cometapi.com/v1/responses \
     --header "Authorization: Bearer $COMETAPI_KEY" \
     --header "content-type: application/json" \
     --data \
'{
    "model": "gpt-5.4-2026-03-05",
    "input": "How much gold would it take to coat the Statue of Liberty in a 1mm layer?",
    "reasoning": {
        "effort": "none"
    }
}'

cURL Code Example

curl https://api.cometapi.com/v1/responses \
     --header "Authorization: Bearer $COMETAPI_KEY" \
     --header "content-type: application/json" \
     --data \
'{
    "model": "gpt-5.4-2026-03-05",
    "input": "How much gold would it take to coat the Statue of Liberty in a 1mm layer?",
    "reasoning": {
        "effort": "none"
    }
}'

Python Code Example

from openai import OpenAI
import os

# Get your CometAPI key from https://api.cometapi.com/console/token, and paste it here
COMETAPI_KEY = os.environ.get("COMETAPI_KEY") or "<YOUR_COMETAPI_KEY>"
BASE_URL = "https://api.cometapi.com/v1"

client = OpenAI(base_url=BASE_URL, api_key=COMETAPI_KEY)

response = client.responses.create(
    model="gpt-5.4-2026-03-05",
    input="How much gold would it take to coat the Statue of Liberty in a 1mm layer?",
    reasoning={"effort": "none"},
)

print(response.output_text)

JavaScript Code Example

import OpenAI from "openai";

// Get your CometAPI key from https://api.cometapi.com/console/token, and paste it here
const COMETAPI_KEY = process.env.COMETAPI_KEY || "<YOUR_COMETAPI_KEY>";
const BASE_URL = "https://api.cometapi.com/v1";

const client = new OpenAI({
    apiKey: COMETAPI_KEY,
    baseURL: BASE_URL,
});

async function main() {
    const response = await client.responses.create({
        model: "gpt-5.4-2026-03-05",
        input: "How much gold would it take to coat the Statue of Liberty in a 1mm layer?",
        reasoning: {
            effort: "none",
        },
    });

    console.log(response.output_text);
}

main();

Uptime

Request success rate over the last 30 days, reflecting the reliability of each model provider. CometAPI monitors all connected providers in real time, 24/7.

RespondLIVE
2165msAvg. Response
UptimeLIVE
100.0%Avg. Uptime

Versions of GPT-5.4

The reason GPT-5.4 has multiple snapshots may include potential factors such as variations in output after updates requiring older snapshots for consistency, providing developers a transition period for adaptation and migration, and different snapshots corresponding to global or regional endpoints to optimize user experience. For detailed differences between versions, please refer to the official documentation.

Model idAvailabilityRequest
gpt-5.4-2026-03-05โœ…Responses and Chat Completions
gpt-5.4โœ…Responses and Chat Completions