o3-mini is a small reasoning model built by OpenAI for technical tasks, focused on mathematics, science, coding, and multi-step logic problems. It offers adjustable reasoning effort and supports function calling, structured outputs, and streaming responses, making it suitable for integration into coding assistants, problem-solving applications, and technical analysis workflows. On this platform, you can choose standard chat, Responses, or managed sessions as needed.
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="o3-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, input/output, and invocation methods before selecting a model.
Native capabilities apply to o3-mini; each endpoint uses messages, input, or question to structure requests, and general service parameters do not mean all are supported by this model.
Learn what o3-mini can bring to your work.
o3-mini focuses on reasoning in mathematics, science, and programming, rather than merely generating fluent text. For problems involving multiple conditions, you can ask it to organize solution steps, explain assumptions, and provide verification methods. When using it, clearly specify variables, units, and constraints to make it easier to obtain verifiable technical answers.
low, medium, and high allow the same model to adapt to tasks of varying complexity. For routine code explanations, start with low; for analyses involving multiple constraints, start with medium; and try high for difficult problems. Increasing intensity is a choice to invest more in reasoning; it does not mean you will get better answers every time, nor does it mean longer responses are required.
Function calling and structured outputs make o3-mini suitable for handling analysis steps in applications. You can define function names and parameters for the model to generate calling recommendations; you can also specify a JSON structure to receive classifications, repair recommendations, or problem-solving results. Tool execution and result validation remain the application's responsibility; operations are not completed automatically merely because call parameters are generated.
Start with specific tasks to find where the model can be effective.
Provide relevant code, error logs, expected behavior, and runtime conditions, and let o3-mini analyze possible causes and produce modification plans and test case recommendations. It is suitable for identifying algorithm boundary issues, missing conditions, and implementation logic problems. Fixes that need to be run in practice should be validated in a test environment rather than deploying generated code directly to production.
Submit the text of the problem, formulas, known conditions, and your own attempted solution process, and let the model provide solution explanations, compare different methods, or check for gaps in derivations. Deliverables can include step-by-step explanations, a list of assumptions, and verification approaches. If the problem comes from an image, convert it to text first and retain necessary descriptions of symbols, units, and geometric relationships.
Organize business rules, configuration conditions, or algorithm requirements into text, and let o3-mini analyze relationships between conditions and generate decision tables, explanations of exception branches, or structured inspection results. It is suitable for technical analysis with clear inputs and acceptance criteria; when using JSON output, define fields and allowed values first, then perform programmatic validation.
Choose based on task complexity, input materials, and expected results.
If you previously used o1-mini for code, math, or science problems, o3-mini is worth considering as a replacement. OpenAI's published evaluations show improvements in answer accuracy and clarity in these areas, along with support for function calling, structured outputs, and developer messages. Test it with your own problem sets, code samples, and formatting requirements before deciding the scope of migration.
For technical problems where you want flexible control over reasoning effort, choose o3-mini; for broader general-knowledge reasoning, consider o1. In some technical evaluations, o3-mini at the medium setting performs close to o1, but this should not be taken to mean they are equivalent for all tasks. o3-mini-high is the high-intensity option in ChatGPT; here, use o3-mini with high rather than switching to that name.
Move from a small-scale task to full integration.
Define objectives, required inputs, and output requirements, using real business examples as a starting point.
Open the playground, confirm the parameters supported by this entry point, then submit a small-scale task to review the results.
Keep the full model ID, use the request format specified in the documentation, and confirm billing rules on the Pricing page.
Understand output quality and capability boundaries before production use.
Answers to common questions about using o3-mini.
It is recommended to start with medium and compare results using real tasks; try low for simple explanations, and high for complex reasoning or difficult coding problems. Chat Completions uses reasoning_effort, while Responses uses reasoning configuration settings. o3-mini's native levels are low, medium, and high; other levels should not be applied to it.
It cannot directly understand screenshots because it does not support vision. First convert the code and errors in the screenshot to text, then submit file snippets, the runtime environment, and expected results. If the screenshot involves layout, graphics, or other information that cannot be fully transcribed, use a model that supports image understanding instead.
Yes, it supports structured outputs. When using Chat Completions, you can select json_schema through response_format and define the field structure, which is suitable for returning diagnostic results, rule classifications, or repair suggestions. Meeting the required structure does not guarantee that the content is correct; applications should still check field values and business conditions.
Existing messages-based conversation code can use Chat Completions; Responses can be chosen for response flows organized with input. To simplify continuous Q&A, you can use AI Chat or AI Chat v2, ask questions through question and receive answer, and include id when saving the conversation to continue the discussion. Different approaches should parse responses separately.
o3-mini-high is the high reasoning effort option provided in ChatGPT when OpenAI released it, not a model ID that needs to be substituted here. When calling it, still select o3-mini and set reasoning effort to high. It is also different from o3; their names or capabilities should not be regarded as completely identical.
Model information · Updated: 2026-10-01. For calling parameters and billing rules, see the API and pricing sections.
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