o3
A dedicated reasoning model that thinks before it answers, strong on math, science and multi-step logic. Favored where careful deliberation matters more than raw speed.
At a glance
Released
Apr 16, 2025
Provider
OpenAI
Context window
200,000 tokens
Max output
100,000 tokens
Modalities
Text + image → text
Knowledge cutoff
March 2024
API model ID
o3License
Proprietary (API only)
Pricing
Standard
Input$2 / 1M
Output$8 / 1M
Batch (50% off)
Input$1 / 1M
Output$4 / 1M
A typical 1,500-token prompt with a 650-token answer costs ≈ $0.008 on the standard tier. Run your own prompt through the calculator →
What's new vs the previous model
Compared witho1 (December 2024)
- Stronger step-by-step reasoning than o1
- Better tool use during thinking
- Improved math and coding accuracy
- Slower due to extended reasoning
- Higher token use on hard problems
Best use cases
Math and science
Works through rigorous quantitative problems carefully.
Complex planning
Breaks down multi-step tasks before acting.
Hard debugging
Reasons through subtle code and logic faults.
Strengths & limitations
Strengths
- Strong deliberate reasoning
- Good at math
- Reliable tool use
- Accurate on hard tasks
Limitations
- Slower responses
- Higher reasoning-token cost
How to test it
- No-code: open chatgpt.com and pick o3 from the model switcher — no API key needed.
- Playground: sign in to the OpenAI Platform and run prompts in the workbench, then copy the generated code.
- API: create a key in the OpenAI Platform and call the API (example below).
curl https://api.openai.com/v1/chat/completions \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "o3",
"messages": [{"role": "user", "content": "Explain tokenization in one paragraph."}]
}'