Existing production systems
Maintain a tested GPT-5.4 integration while evaluating a newer model against real quality and cost targets.
Independent tool overview
GPT 5.4 at a glance
GPT-5.4 is a supported previous-generation OpenAI reasoning model for coding, professional documents and tool-using agents, with a 1.05-million-token API context window, 128,000-token maximum output and native computer use.
Visit the official GPT 5.4 site ↗
- API model
- gpt-5.4
- Pinned snapshot
- gpt-5.4-2026-03-05
- Context window
- 1,050,000 tokens
- Maximum output
- 128,000 tokens
- Knowledge cutoff
- August 31, 2025
- Current position
- Supported API model; superseded by newer GPT families
Overview
What GPT 5.4 is
OpenAI launched GPT-5.4 in March 2026 across ChatGPT, the API and Codex. It combined general reasoning with coding capabilities from GPT-5.3-Codex and added native computer control, stronger tool search and improved work on spreadsheets, presentations and documents.
The API model remains listed as gpt-5.4, with a dated gpt-5.4-2026-03-05 snapshot for teams that need behavior stability. It accepts text and image inputs, returns text, and supports structured outputs, web and file search, code execution, hosted shell, MCP, skills and computer use.
GPT-5.4 is no longer OpenAI's recommended starting point for new systems; the current guidance points developers to the GPT-5.6 family. Existing applications should compare migration quality, latency and total token cost on representative tasks rather than switching solely on benchmark claims.
Use cases
Who GPT 5.4 is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Long-context professional work
Analyze large document sets or codebases when the 1.05M context window is worth its special pricing.
Computer-using agents
Build screenshot, mouse, keyboard and browser workflows with native computer-use support and explicit confirmation policies.
Capabilities
Core GPT 5.4 features
Configurable reasoning
The API supports none, low, medium, high and xhigh reasoning effort to balance latency, cost and task quality.
Native computer use
GPT-5.4 can issue mouse and keyboard actions from screenshots or generate code for browser automation libraries.
Large context and output
The API supports up to 1.05M tokens of context and 128K output tokens, with a surcharge above the standard long-context threshold.
Professional artifact work
The launch emphasized spreadsheets, presentations, documents, legal analysis and other multi-step professional deliverables.
Broad tool support
Responses API tools include web search, file search, image generation, code interpreter, hosted shell, apply patch, skills, MCP and tool search.
High-fidelity image input
The original detail level supports full-fidelity perception up to 10.24M total pixels or a 6,000-pixel maximum dimension, whichever limit is reached first.
Pinned model snapshot
Teams can select the March 5, 2026 snapshot instead of relying on an alias when reproducibility matters.
Process
How the GPT 5.4 workflow works
Step 1
Decide whether to keep it
Use GPT-5.4 when an existing production evaluation supports it; start new projects with current OpenAI model guidance.
Step 2
Choose a reasoning baseline
Set effort intentionally and measure task success, latency and token use instead of defaulting to xhigh.
Step 3
Control tools and approvals
Give agents the minimum required tools, isolate computer-use environments and require confirmation for consequential actions.
Step 4
Watch the long-context threshold
Retrieve only relevant material where possible because inputs above 272K tokens change pricing for the entire request.
Step 5
Run a migration evaluation
Compare GPT-5.4 with GPT-5.6 at the same reasoning effort and one level lower, using a pinned task set and real total cost.
Cost
GPT 5.4 pricing and free plan
Standard GPT-5.4 costs $2.50 per million input tokens, $0.25 cached input and $15 output. GPT-5.4 Pro costs $30 input and $180 output. Long inputs above 272K tokens trigger higher rates for the full request.
GPT-5.4 standard API
$2.50 input / $15 output per 1M tokens
General reasoning, coding, agent and professional-work model.
- $0.25 per 1M cached input tokens
- Batch and Flex processing are half the standard rate
- Priority processing is twice the standard rate
- Regional processing carries a 10% uplift
GPT-5.4 Pro API
$30 input / $180 output per 1M tokens
Higher-compute version for difficult quality-first tasks.
- No cached-input price listed
- Substantially higher latency and cost should be expected
- Use only when measured quality gains justify the difference
Long-context adjustment
Above 272K input tokens
Special pricing applies to the full request once input crosses the threshold.
- 2x input-token rate
- 1.5x output-token rate
- Applies to standard, Batch and Flex processing
- Retrieve and compact context before sending very large prompts
Pricing checked . Check current pricing at the source ↗
Assessment
GPT 5.4 strengths and limitations
Where it stands out
- Large context and output limits for complex, long-horizon workflows
- Native computer use plus broad hosted-tool support
- Pinned snapshot supports reproducible production deployments
- Strong fit for code, spreadsheets, presentations and documents
- Cached-input, Batch and Flex pricing can reduce recurring workload cost
What to consider
- It is no longer OpenAI's current recommended flagship for new applications
- Inputs above 272K tokens increase both input and output rates for the entire request
- The August 2025 knowledge cutoff requires search or supplied sources for newer facts
- Text and image input are supported, but direct audio and video input are not
- Computer use remains fallible and requires isolation, approvals and post-action verification
- GPT-5.4 Pro is dramatically more expensive than the standard model
Compare
GPT 5.4 alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Business Operations
ChatGPT
Choose current ChatGPT when you want OpenAI to route everyday work through its newest supported consumer models.
Explore ChatGPT →Consumer
GPT 5.5
Compare GPT-5.5 when maintaining an intermediate OpenAI migration path, though new API work should also test GPT-5.6.
Explore GPT 5.5 →Consumer
Gemini 3.5 Flash
Choose Gemini 3.5 Flash when lower-latency, lower-cost multimodal throughput matters more than staying within OpenAI's stack.
Explore Gemini 3.5 Flash →Questions
GPT 5.4 FAQs
Is GPT-5.4 still available?
Yes. OpenAI's current API catalog still lists gpt-5.4 and the pinned gpt-5.4-2026-03-05 snapshot. It is a supported previous-generation model rather than the recommended starting point for new builds.
How much does the GPT-5.4 API cost?
Standard pricing is $2.50 per million input tokens, $0.25 cached input and $15 output. Regional, Priority and long-context adjustments can increase the effective rate.
What is GPT-5.4's context window?
The API model page lists a 1,050,000-token context window and a 128,000-token maximum output.
Does GPT-5.4 support computer use?
Yes. It supports native computer use through screenshot-based mouse and keyboard actions as well as code-driven browser automation. Developers still need permission boundaries and human confirmation for consequential actions.
What happens above 272K input tokens?
For GPT-5.4 and GPT-5.4 Pro, OpenAI charges 2x the input rate and 1.5x the output rate for the full request when input exceeds 272K tokens under standard, Batch or Flex processing.
Should I migrate from GPT-5.4 to GPT-5.6?
For new work, OpenAI recommends the GPT-5.6 family. For an existing application, preserve GPT-5.4 as the baseline, test GPT-5.6 at the same reasoning effort and one level lower, and compare quality, latency and total cost on representative tasks.
Bottom line
Our GPT 5.4 verdict
GPT-5.4 remains useful where it is already evaluated, especially for long-context and computer-use systems. New deployments should treat it as a migration baseline and validate GPT-5.6 before committing to an older model generation.
Visit GPT 5.4 website ↗Similar Consumer Tools
View all Consumer AI tools →
Gemini 3.1 Flash-Lite
Gemini 3.1 Flash-Lite - Google's fastest, cheapest Gemini 3 model for high-volume dev workloads
Copilot Cowork
Copilot Cowork - Microsoft's Anthropic-powered AI for running multi-step tasks across M365 apps
GPT-5.3 Instant
GPT-5.3 Instant - OAI's ChatGPT default model update with fewer refusals and less hallucinations
Nemotron 3 Super
Nemotron 3 Super - NVIDIA's open 120B reasoning model with 1M token context for agentic workflows
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.