Back to models
Chat Models

OpenAI

GPT-4.1

OpenAI non-reasoning model for instruction following, coding, and long documents, with a 1,047,576-token context window.

Long contextCoding
Open AppBrowse All Models

At a glance

Know the model before you prompt.

OpenAI API specifications
Context window
1,047,576 tokens
Maximum output
32,768 tokens
Inputs → output
Text + Images → Text
Knowledge cutoff
June 1, 2024

API model ID: gpt-4.1

Capabilities & boundaries

What it supports. Where the limits are.

Tool support requires the appropriate API integration; a supported tool is not automatically active in every chat.

Supported API features and tools

  • Streaming responses
  • Function calling and structured outputs
  • Predicted outputs and prompt caching
  • Fine-tuning capability, subject to current platform access restrictions
  • Web search, file search, image generation tools, code interpreter, and MCP

Before you choose

  • No native audio or video support; native output is text.
  • This is a non-reasoning model, without configurable reasoning effort.
  • Fine-tuning is restricted: organizations that have never fine-tuned cannot create new jobs, and inactive organizations were restricted July 2, 2026. New jobs end January 6, 2027 for remaining active customers; existing fine-tuned inference continues until the base model is deprecated.
  • A 1,047,576-token context does not imply equally large answers: maximum output is 32,768 tokens. Verify retrieval and coverage on long inputs.

Non-reasoning model

GPT-4.1 answers without a separate reasoning step. There is no low/medium/high effort selector or reasoning default to configure here. Improve results with clear requirements, relevant evidence, examples, and a checkable output format; compare against a reasoning model for tasks that need deeper deliberation.

  • Chat Completions, Responses, and Batch are supported. The listed built-in tools require the appropriate Responses integration.
  • The Assistants API retired August 26, 2026. Do not build a new integration around the historical Assistants listing on this model’s reference page.
  • The documented snapshot is gpt-4.1-2025-04-14. API rate limits depend on usage tier, with separate long-context limits; the free API tier is not supported.
  • Fine-tuning access changed May 7 and July 2, 2026; consult the deprecation notice for organization-specific restrictions rather than treating a supported capability as open enrollment.

Put it to work

Start with a more useful prompt.

Original examples from EZ Ai Assist. Adapt these to your task and the features available in your workspace.

Workflow 01

Apply an exact code-edit contract

Use explicit invariants and an allowed scope to constrain a small change.

Update the supplied function according to this change contract. Preserve its public signature, error messages, and behavior outside the listed cases. Explain each proposed edit by referencing a requirement, and include tests for the changed cases plus one unchanged case. Do not edit unrelated files, introduce dependencies, or claim the tests passed without execution evidence.

Workflow 02

Build a source-indexed document map

Organize a long input into references that a reviewer can verify.

Create a topic map of the supplied documents using their existing section titles and page numbers. For each requested topic, list the relevant passages, a one-sentence summary, and any contradictions between sources. Mark topics with no evidence as not found. Do not infer missing policy or merge conflicting requirements into a single unsupported statement.

Workflow 03

Transform text to a house style

Separate formatting and wording changes from factual changes.

Rewrite this draft to follow the supplied house-style rules while preserving every factual claim, number, qualification, and named entity. Return the revised draft, then a short checklist of rules applied and any source ambiguities you could not resolve. Do not add examples or strengthen claims beyond the original, and flag conflicting style requirements for review.

Developer reference

OpenAI API pricing

These are OpenAI API reference prices, not EZ Ai Assist subscription prices.

View EZ Ai Assist plans
Standard processing · USD per 1,000,000 tokens
Token typePrice
Input$2.00
Cached input$0.50
Output$8.00
  • These are Standard base-model GPT-4.1 rates. Fine-tuned model training and inference have separate prices and access restrictions.
  • Cached pricing applies only to eligible cached tokens. The reviewed pricing table does not list a separate long-context surcharge; separate rate limits are not separate token prices.
  • Tools, predicted-output usage rules, and processing tiers can affect billing. Check the current documentation before estimating cost.

Common questions

A few things worth knowing.

Is GPT-4.1 a reasoning model?

No. OpenAI describes it as non-reasoning, without a separate reasoning step. Do not send an effort setting copied from GPT-5 or o3. Clear instructions and validation still matter, and a reasoning model may perform differently on complex work.

How large are its context and output limits?

The model reference lists a 1,047,576-token context window and a 32,768-token maximum output. Long input capacity does not guarantee complete retrieval or equally long answers. Use section references and test coverage on the documents you actually need.

Can new customers fine-tune GPT-4.1?

The model supports fine-tuning as a capability, but platform access is restricted. Since May 7, 2026, organizations that never fine-tuned cannot start; July 2 restrictions cover inactive organizations. Remaining active customers lose new-job creation January 6, 2027. Check the current notice for your organization’s eligibility.

Will existing fine-tuned models stop in January 2027?

The notice distinguishes creating new training jobs from inference. Existing fine-tuned model inference continues until the underlying base model is deprecated. Do not interpret the January 6, 2027 new-job deadline as an announced GPT-4.1 base-model shutdown.

Should I use the Assistants API listed on its model page?

No. The platform’s Assistants API retired August 26, 2026 even though the model reference retains a historical support listing. Use current supported endpoints and migration guidance rather than relying on that older endpoint row.

What are predicted outputs and structured outputs?

Predicted outputs can help supported editing workflows when much of the intended output is already known; they have their own API and billing rules. Structured outputs constrain format, not truth. Neither feature should be assumed available through a plain prompt or every EZ Ai Assist interface.

Do its API prices or tools describe my subscription?

No. This guide lists provider API capabilities and base-model prices. EZ Ai Assist subscriptions, available inputs, tools, and model access are separate. Adapt the examples to the current app and verify important results before use.

Check the source

Official documentation

Specifications and API prices checked on . Example prompts and workflow advice are editorial guidance from EZ Ai Assist.

Same provider

More from OpenAI

OpenAI

GPT-6.1 Sol

GPT-6.1 Sol is OpenAI’s reasoning model for complex coding and professional work, with image input and a 1.05M-token context window.

OpenAI

GPT-6 Sol

GPT-6 Sol is an OpenAI model for complex coding and tool-assisted workflows, with image input and configurable reasoning.

OpenAI

GPT-6 Luna

GPT-6 Luna is an efficient OpenAI model for focused, high-volume tasks, with image input, structured outputs, and configurable reasoning.

OpenAI

GPT-6 Astra

GPT-6 Astra is an OpenAI model for complex reasoning, coding, and research, with image input and a 1.05M-token context window.