ChatImage

Model developer guide

OpenAI

39 website model guides

Choose a model for the job: reasoning and writing with GPT, or creating and editing visuals with GPT Image. This guide connects practical starting points to model-specific API details, without treating every OpenAI feature as an EZ Ai Assist control.

ReasoningCodingImage generation

Practical guidance

Where should I start?

Editorial starting points, not a performance ranking. Check current app access before choosing a model.

Complex writing or code review

Start an evaluation with GPT-6.1 Sol for multi-step work that needs reasoning over supplied evidence. OpenAI positions it as a balance of capability and cost.

Tradeoff: More reasoning is not a substitute for checking citations, running tests or measuring response time on your task.

GPT-6.1 Sol guide

Focused, repeatable text tasks

Consider GPT-6 Luna for a repeatable task such as sorting short messages or extracting a few fields. Its official positioning is focused, high-volume work.

Tradeoff: Evaluate accuracy on difficult examples before choosing efficiency over a stronger result on complex inputs.

GPT-6 Luna guide

Precise edits or everyday visuals

OpenAI recommends GPT Image 2.5 Sunburst where editing precision matters. Flare is the companion option positioned for fast, everyday image generation; both have guides below.

Tradeoff: Inspect text, product details and reference fidelity. Provider positioning is not a benchmark measured inside this app.

Sunburst image guide

Technical details

What does the provider document?

Official documentation describes the named model or gateway—not every feature available in this app.

Text models: context is a budget

GPT-6.1 Sol and GPT-6 Luna each document a 1,050,000-token context window, a 922,000-token maximum input and up to 128,000 output tokens. These are model-specific API ceilings, not the size of a file you can necessarily upload in EZ Ai Assist. Both accept text and images and produce text.

Reasoning and tools depend on the endpoint

Sol supports low, medium, high, xhigh and max reasoning; none and minimal are unsupported. Luna also supports none. Sol tool calling requires the Responses API; Luna supports Chat Completions function calling only with reasoning_effort set to none. Do not transfer one model’s settings to every GPT model.

Image models have a separate contract

Sunburst and Flare take text/image input and produce images through image generation/edit endpoints or the Responses image-generation tool. Both list low, medium, high, xhigh, max and auto quality. Their pinned snapshots end in -2026-09-08. A chat model’s image-input support alone does not mean it outputs images.

Complete website coverage

Explore all 39 model guides

Every existing catalog entry for this provider is linked below. Status labels come from the individual website guides; catalog coverage is not a guarantee of current app access. Historical and snapshot pages remain references, not extra active models.

Practical guidance

From a brief to a useful result

Original prompt examples, not generated-output demonstrations. They have not been generation-tested.

Workflow 1

Turn messy notes into a decision brief

Provide a short, non-sensitive set of notes and your decision criteria. Keep the first trial small enough to check line by line.

Using only the notes below, compare the options against implementation effort, customer impact and uncertainty. Cite the supporting note for every factual claim. Separate facts from assumptions. Recommend the next small test, not a final decision where evidence is missing. Finish with three questions for the team.

Check the result: Verify each cited note exists, the recommendation follows your criteria, and invented costs or claims have not slipped in.

Use the Sol guide

Workflow 2

Make a controlled product-image revision

If reference-image editing is available in your app session, attach a product photo you have permission to use. Specify both the change and what must remain untouched.

Using the attached bottle photo, replace only the background with a pale sandstone surface and soft morning light. Preserve the bottle shape, cap, label wording, logo position and camera angle. Keep the shadow believable. Do not invent product claims, add accessories or change the packaging.

Check the result: Compare the label and geometry with the original at full size. Reject altered branding rather than assuming a polished image is an accurate edit.

Use the Sunburst guide

Technical details

What changes inside EZ Ai Assist?

Official OpenAI references were checked for the capabilities highlighted here. App image controls are reported by the October 8, 2026 owner audit in our image guides; chat request parameters were not independently inspected. A catalog page is not a live availability check.

Website catalog

What is available to choose?

The complete list below is the website’s existing OpenAI coverage, including historical references. It does not establish that every entry remains selectable, enabled for your account or operational in the app.

Owner audit · October 8

What controls are described in the app?

The Sunburst and Flare guides report quality, format and background controls plus up to four reference images. Their audited upstream model IDs are gpt-image-2.5-sunburst-2026-09-08 and gpt-image-2.5-flare-2026-09-08. This is image-route evidence, not a promise that every OpenAI model shares these controls.

Unverified in this guide

What still needs an app-level check?

The current chat endpoint, reasoning value, tool configuration, upload limits and final forwarded request were not inspected. OpenAI’s large context windows, hosted tools and service tiers must not be read as confirmed EZ Ai Assist features.

Before you rely on a result

Limits, costs and tradeoffs

  • A good-looking answer may still contain factual or coding errors. Review the output against supplied evidence and run appropriate checks before using it.
  • Provider API pricing, subscription pricing and usage shown by EZ Ai Assist are different things. Compare the exact model, input/output units and processing options; do not assume a flat cost per answer or image.
  • Older model and snapshot guides remain useful references, but are not recommendations to start new work on a retired endpoint. Check the linked guide’s lifecycle notice and official documentation.

Provider references and picker estimates are not EZ Ai Assist customer charges. Read the website pricing explanation.

Compare by task

Other paths to explore

Anthropic

Compare another model family for the same writing or coding brief, using your own acceptance criteria.

Google

Explore another catalog for text and visual workflows; compare exact model inputs and app controls.

FLUX

Explore image-focused models when your task is visual creation or editing rather than a text response.

Common questions

OpenAI, without the assumptions

Is using OpenAI here the same as using ChatGPT?

No. This is EZ Ai Assist’s model guide, not the ChatGPT product. The model, interface, tools, limits and billing arrangement must be checked separately; a feature described for one product is not automatically available in the other.

Why are there more guides than recommended starting points?

The full list preserves existing model URLs and includes older models and snapshot references. The starting points are editorial suggestions for evaluation, not a claim that all catalog entries are active app choices.

Does a million-token context mean I can upload anything?

No. The cited figures are API ceilings for the named models. App upload limits, file processing, tool use, output budget and cost can impose other constraints. Start with relevant excerpts and check what your app session supports.

New to the terminology?

Official documentation

Sources & review notes

Highlighted provider contracts reviewed . Individual model guides carry their own review dates. Sources can change; these pages do not perform live health checks.

Official OpenAI references were checked for the capabilities highlighted here. App image controls are reported by the October 8, 2026 owner audit in our image guides; chat request parameters were not independently inspected. A catalog page is not a live availability check.