o3
o3 is a well-rounded and powerful model across domains. It sets a new standard for math, science, coding, and visual reasoning tasks.
o3Released 16 Apr 2025 (Apr 2025)Pricing
List prices at OpenAI, in US dollars per million tokens.
| Tier | Input / 1M | Output / 1M | |
|---|---|---|---|
| Standard | $2.00 | $8.00 | |
| Cached input 75% off input | $0.50 | 75% off input | |
| Batch 50% off | $1.00 | $4.00 | 50% off |
| Flex 50% off | $1.00 | $4.00 | 50% off |
| Priority 75% more | $3.50 | $14.00 | 75% more |
| Web search $10.00 per 1,000 searches | $10.00 per 1,000 searches |
What it costs a month
At OpenAI list prices, 30 days.
- Chat assistant5,000 requests a day, 2K in / 0.5K out, cached$1,065
- RAG over documents2,000 requests a day, 8K in / 0.6K out, cached$888
- Coding agent1,000 requests a day, 40K in / 2K out, cached$1,620
- Overnight batch20,000 requests a day, 4K in / 0.4K out, batch$3,360
Where to run it
3 hosts serve o3. The comparison is on blended price, 3 parts input to 1 part output.
| Host | In / out per 1M | Model id | |
|---|---|---|---|
OpenAILab API o3 | $2.00 / $8.00 Retires 11 Dec 2026 | o3 | Retires 11 Dec 2026 |
Azure OpenAICloud o3 | $2.00 / $8.00 Retires 19 Nov 2026 | o3 | Retires 19 Nov 2026 |
Vercel AI GatewayGateway o3 | $2.00 / $8.00 | o3 |
Show 6 more listings: regions, snapshots and variantsHide regional listings
| OpenAI | |||||||
| 2025-04-16 o3-2025-04-16 | $2.00 / $8.00 Retires 11 Dec 2026 | o3-2025-04-16 | Retires 11 Dec 2026 | ||||
| Azure OpenAI | |||||||
| 2025-04-16 o3-2025-04-16 | $2.00 / $8.00 Retires 19 Nov 2026 | o3-2025-04-16 | Retires 19 Nov 2026 | ||||
| us o3 | $2.20 / $8.80 +10% vs headline Retires 19 Nov 2026 | o3 | Retires 19 Nov 2026 | ||||
| eu o3 | $2.20 / $8.80 +10% vs headline Retires 19 Nov 2026 | o3 | Retires 19 Nov 2026 | ||||
| us · 2025-04-16 o3-2025-04-16 | $2.20 / $8.80 +10% vs headline Retires 19 Nov 2026 | o3-2025-04-16 | Retires 19 Nov 2026 | ||||
| eu · 2025-04-16 o3-2025-04-16 | $2.20 / $8.80 +10% vs headline Retires 19 Nov 2026 | o3-2025-04-16 | Retires 19 Nov 2026 | ||||
Capabilities
Core
- Streaming
- Function calling
- JSON mode
- Structured outputs
Multimodal
- Vision
- Image input
- Audio
- Image generation
- Video
Advanced
- Reasoning
- Tool calling
- MCP
- Prompt caching
- Batch API
- Embeddings
- Takes in
- ImagesTextPDFs
- Produces
- Text
- Features
- PDF inputPrompt cachingBuilt-in web search
Cheaper alternatives
Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.
Same model, cheaper host
No host sells it for less than OpenAI.
Smaller OpenAI models
- 43% less
- 52% less
- 87% less
Other frontier labs
- 14% less
- 14% less
- 94% less
- 57% less
Open-weight picks
- 36% less
- 51% less
- 80% less
- 85% less
Benchmarks
Scores published on the model card.
- Aider Polyglot81.3%
Model line
Newer and older versions, at each lab's own price.
- Apr 2025o3Latest$2.00 / $8.00
- Dec 2024o1$15.00 / $60.00
Call it
Using OpenAI's API.
from openai import OpenAI
client = OpenAI() # reads OPENAI_API_KEY
response = client.responses.create(model="o3", input="Hello")
print(response.output_text)curl https://api.openai.com/v1/responses \
-H "Authorization: Bearer $OPENAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "o3", "input": "Hello"}'from litellm import completion
response = completion(
model="openai/o3",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)Quick answers
How much does o3 cost?
On OpenAI, o3 costs $2.00 per million input tokens and $8.00 per million output tokens. Cached input costs $0.50 per million tokens.
Which APIs offer o3?
o3 runs on 3 APIs: OpenAI, Azure OpenAI and Vercel AI Gateway.
What is o3's context window?
200K tokens, with up to 100K tokens in one reply.
Is o3 being retired?
Azure OpenAI retires it on 19 Nov 2026; OpenAI on 11 Dec 2026. 1 host keeps serving it.
Is there a newer version of o3?
No. o3 is the newest model in its line.
What are cheaper alternatives to o3?
Models that cost less and keep most of what it can do: GPT-6.1 Terra on OpenAI (43% less), Gemini 4 Hydrogen on Google AI Studio (14% less), Grok 4.7 on xAI (14% less) and Qwen4 Max on Alibaba DashScope (36% less). Savings are on a mix of 3 input tokens to 1 output token.
When was o3 released?
OpenAI released it on 16 Apr 2025.
What is o3's API model id?
Use o3 with OpenAI.
Sources
Raw catalog entry for o3 on OpenAI
{
"cache_read_input_token_cost": 5e-7,
"cache_read_input_token_cost_flex": 2.5e-7,
"cache_read_input_token_cost_priority": 8.75e-7,
"deprecation_date": "2026-12-11",
"deprecation_source": "https://developers.openai.com/api/docs/deprecations",
"successor_model": "gpt-6.1-sol",
"input_cost_per_token": 0.000002,
"input_cost_per_token_batches": 0.000001,
"input_cost_per_token_flex": 0.000001,
"input_cost_per_token_priority": 0.0000035,
"litellm_provider": "openai",
"max_input_tokens": 200000,
"max_output_tokens": 100000,
"mode": "chat",
"output_cost_per_token": 0.000008,
"fedramp_uplift_multiplier": 1.1,
"output_cost_per_token_batches": 0.000004,
"output_cost_per_token_flex": 0.000004,
"output_cost_per_token_priority": 0.000014,
"reasoning_contexts": [
"auto",
"current_turn"
],
"reasoning_modes": [
"standard"
],
"search_context_cost_per_query": {
"search_context_size_high": 0.01,
"search_context_size_low": 0.01,
"search_context_size_medium": 0.01
},
"source": "https://developers.openai.com/api/docs/pricing",
"supported_endpoints": [
"/v1/responses",
"/v1/chat/completions",
"/v1/completions",
"/v1/batch"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"text"
],
"supports_function_calling": true,
"supports_parallel_function_calling": false,
"supports_pdf_input": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true
}