o4-mini is a small model from OpenAI built for efficient reasoning, suited to tasks that require analytical processes, code evaluation, and image-text understanding. Its focus is not simple expansion, but breaking down problems based on conditions, comparing solutions, and using configured tools for verification. Applications can integrate it using the public request format in this page's API section.
Point the Base URL to api.acedata.cloud, configure your platform API key and the model ID below, and use your compatible SDK or client.
import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["ACEDATACLOUD_API_KEY"],
base_url="https://api.acedata.cloud/v1",
)
response = client.responses.create(
model="o4-mini",
input="Hello!",
)
print(response.output_text)Choose an available protocol for this model. OpenAI SDK uses a Base URL ending in /v1; Anthropic SDK uses the root URL. See each guide for protocol-specific parameters, tools and response formats.
Clarify capacity, inputs and outputs, and invocation methods before selecting a model.
o4-mini's mathematical, coding, and visual reasoning are model capabilities. Applications pass relevant history, tool results, and output budgets through Chat Completions or Responses; actual operations are completed by the execution environment.
Learn what o4-mini can bring to your work.
o4-mini excels at math and logic tasks, making it suitable for turning problem conditions into solution steps, checking candidate answers, and then explaining key conclusions. When using it, you can ask it to list assumptions, edge cases, and validation methods rather than only providing the final number. For complex calculations, connecting computational tools can more clearly divide analysis from numerical verification.
Programming is one of o4-mini's main strengths. After providing code snippets, error messages, and expected behavior, you can ask it to analyze the problem, compare repair approaches, and organize testing recommendations. It is also suitable for assisting with implementation discussions in data science tasks; whether code can run still needs to be verified in the actual environment, as model responses do not automatically execute programs.
o4-mini can analyze problems by combining images and text, making it suitable for reading textbook diagrams, whiteboard notes, and hand-drawn sketches, then explaining relationships, conditions, or points of uncertainty. It is not only used to describe images, but can also participate in solving problems based on diagrams. When submitting images, include a specific question and indicate the area of interest, which is usually more effective than broadly requesting analysis.
Start with specific tasks to find where the model can be effective.
Enter a photo of the problem, the known conditions, and your own attempt, and have o4-mini organize the problem statement, identify breaks in the derivation, and deliver a checkable solution explanation. Suitable for teaching assistance and practice feedback. If symbols in the image are unclear, you can also provide the conditions in text to prevent recognition errors from being amplified throughout subsequent reasoning.
Submit the relevant functions, error logs, and input samples, and ask o4-mini to provide cause analysis, modification plans, and test cases covering edge cases. Deliverables can include patch suggestions and a testing checklist. Limit the task to a specific module and specify the language version, dependencies, and behavioral constraints to receive answers that better fit the project.
When using Chat Completions, put relevant history into messages; when using Responses, organize input and related conversation content according to the documentation. In each round, provide the latest materials, revision goals, and key constraints; for longer tasks, retain interim summaries and a final version that can be checked independently.
Choose based on task complexity, input materials, and expected results.
When tasks focus on math, code, or image and text analysis and require frequent use, o4-mini is an efficiency-focused choice. o3 is better suited to complex, multifaceted reasoning, especially for questions with non-obvious answers that require in-depth comparison of hypotheses. Start by evaluating o4-mini's results on representative examples, then decide which difficult tasks need to be reviewed by o3.
Compared with o3-mini, o4-mini expands performance on non-STEM and data science tasks in official evaluations, while also being suitable for incorporating visual materials into reasoning. If an existing math or coding assistant is starting to handle diagrams, screenshots, and business analysis, prioritize evaluating it. o4-mini-high is an option with greater reasoning effort and should not be treated as another independent model variant to call here.
Based on o4-mini's characteristics, first validate a small task whose results can be checked.
You can ask directly: Based on the problem and the clear diagram, list the known conditions, provide a solution method and verification steps. Do not infer lengths not labeled in the diagram, and finally check whether the answer satisfies all conditions.
Submit diagrams and textual conditions together; check assumptions, calculations, and counterexamples rather than looking only at the final answer.
Use the full model ID o4-mini, first confirm the public request format and available parameters on the API page, then connect your application. Keep result parsing, exception handling, and relevant evidence, and use the same set of real samples to evaluate whether it is suitable for continued use.
Before formal use, understand the output quality and scope of capabilities.
Answers to common questions about using o4-mini.
It can be used for conversation, but it is more valuable for tasks that require derivation, code analysis, or image-and-text judgment. If the need is only a brief rewrite, a reasoning model may not be necessary; for problems with many conditions or requiring validation of an approach, o4-mini is a better fit.
When using Chat Completions, combine text and image_url content blocks in the content of messages, with text describing the problem to solve. It is recommended to have it first confirm the conditions in the image, then provide a solution; formulas or annotations that are unclear in the image should be entered separately as text.
It will not execute Python code automatically simply because it generated it. You need to configure a tool capable of running code and complete the call and result return. When no execution tool is connected, you can have it provide code and validation steps, but numerical results and execution results should be considered separately.
Use Chat Completions or Responses and provide the complete model ID. Chat Completions uses messages and choices, while Responses uses input and the corresponding response structure; handle history management, streaming events, and tool parameters separately according to the chosen interface, without mixing the two formats.
o4-mini-high is an option in ChatGPT associated with greater reasoning effort; it does not mean you need to use a different model ID here. Continue to choose o4-mini for calls; reasoning configuration should match the interface in use, and quality and response experience should be compared through actual tasks.
Choose a development approach or application suited to this model and follow its setup guide.
↗OpenAI Chat Completions API
SDK installation, authentication and examples↗OpenAI Responses API
SDK installation, authentication and examples
↗Using Ace Data Cloud with Cherry Studio
Chat client setup · Base URL, API key and model
↗Using Ace Data Cloud with ChatBox
Chat client setup · Base URL, API key and modelInput
$0.22per 1M tokens
Official $1.10
Output
$0.88per 1M tokens
Official $4.40
You pay$13,584for 142,800 credits · $0.0951 / credits46% OFF
You're billed in credits. Pick a top-up tier to see its per-credit rate — the larger the top-up, the cheaper each credit.
This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs