o3
by OpenAI
o3 pricing, context window and specs
o3 is an OpenAI model with a 200,000 token context window. It costs $2.00 per million input tokens and $8.00 per million output tokens, and supports reasoning, tool calling and structured output.
o3 specs
Provider
OpenAI
Model ID
o3
Input price
$2.00 per 1M tokens
Cached input price
$0.50 per 1M tokens
Output price
$8.00 per 1M tokens
Context window
200,000 tokens
Modalities
Text in, text out
Tool calling
โ Supported
Structured output
โ Supported
Reasoning
โ Supported
Vision
Not supported
Prices checked on
When to use o3
Pick o3 as an everyday model when you need solid quality without paying top-tier prices.
Drafting and analysis. Writing, summarising and answering questions over your own data.
Problems that need thinking. It can reason through a task before it answers.
Agents that use tools. It supports tool calling and structured output, so it can act and hand back clean JSON.
Large inputs. The 200,000 token context window fits long documents and transcripts in one request.
If you need more capability, GPT 5.4 costs $2.50 per million input tokens.
Cost calculator
What o3 costs at your volume
Pick a workload or move the sliders. The estimate uses the prices in the table above.
Templates
Templates you can run with o3
Ready-made workflows and agents. Copy one, pick o3 in the AI step and run it.
Prompt
Lead Fit Scoring & Rationale Builder
A reusable sales prompt template for small teams and founders to evaluate inbound or outbound leads, explain the scoring rationale, surface missing qualification information, and recommend the best next sales action.

Fetch Hive
Build with it
Use o3 in Fetch Hive
Fetch Hive is an AI agent builder. Pick this model in any prompt, workflow or agent step, or call it from your own code with one API request.
Call o3 through the AI Gateway
1
Create a prompt and pick the model
Write your prompt in Fetch Hive and choose this model from the model library. Add a fallback model if you want a backup.
2
Deploy the prompt
Deploying gives the prompt a deployment name and a variant that your code can call.
3
Send one POST request
Call the invoke endpoint with your API key. Usage is charged to your AI wallet at the provider list price, and each call uses one task.
Compare
Compare o3 with similar models
List prices per 1M tokens and context window, next to the models people most often weigh it against.
GPT 4.1
GPT 5.6 Terra
Claude Sonnet 5
FAQ




