Models

Reasoning Tiers

Most model families are available in reasoning variants — High, Medium, Low, and in some cases Minimal. These tiers control how much thinking the model does before responding.

Tier	Behavior
High	Deepest reasoning; slower but most accurate. Best for complex, multi-step problems.
Medium	Balanced reasoning and speed. Good for most complex tasks.
Low	Lightweight reasoning. Suitable for everyday problems where speed matters.
Minimal	Fastest available response. Best for simple, high-volume agents.
(none)	No reasoning overhead. Fastest response time for straightforward tasks.

Text Models

gpt-5.1

Variant	Input	Output
High	272k	128k
Medium	272k	128k
Low	272k	128k
(none)	272k	128k

gpt-5-mini

Variant	Input	Output
High	136k	64k
Medium	136k	64k
Low	136k	64k
Minimal	136k	64k

gpt-5.2

Variant	Input	Output
High	272k	128k
Medium	272k	128k
Low	272k	128k
(none)	272k	128k

gpt-5.3-codex

Variant	Input	Output
High	272k	128k
Medium	272k	128k
Low	272k	128k
(none)	272k	128k

gpt-5.4

Variant	Input	Output
High	872k	128k
Medium	872k	128k
Low	872k	128k
(none)	872k	128k

gpt-5.4-mini

Variant	Input	Output
High	272k	128k
Medium	272k	128k
Low	272k	128k
(none)	272k	128k

gpt-5.4-nano

Variant	Input	Output
High	272k	128k
Medium	272k	128k
Low	272k	128k
(none)	272k	128k

gpt-5.6-sol

Variant	Input	Output
High	872k	128k
Medium	872k	128k
Low	872k	128k
(none)	872k	128k

gpt-5.6-terra

Variant	Input	Output
High	872k	128k
Medium	872k	128k
Low	872k	128k
(none)	872k	128k

gpt-5.6-luna

Variant	Input	Output
High	872k	128k
Medium	872k	128k
Low	872k	128k
(none)	872k	128k

Realtime Voice Models

gpt-realtime

Variant	Input	Output
(none)	28,672	8,191

gpt-realtime-1.5

Variant	Input	Output
(none)	32,000	4,096

gpt-realtime-2

Variant	Input	Output
High	128,000	32,000
Medium	128,000	32,000
Low	128,000	32,000
Minimal	128,000	32,000

gpt-realtime-2.1

Variant	Input	Output
High	128,000	32,000
Medium	128,000	32,000
Low	128,000	32,000
Minimal	128,000	32,000

Available Voices

Every Realtime Voice model can be paired with any of the following synthesized voices:

Voice	Description
Alloy	Neutral and balanced voice with natural, smooth delivery for general use
Ash	Clear and precise voice that conveys sincerity for professional settings
Ballad	Melodic and smooth voice, perfect for creative content and storytelling
Coral	Warm and friendly voice, ideal for customer service and conversations
Echo	Articulate and resonant voice, excellent for professional presentations
Sage	Calm and thoughtful voice conveying wisdom for educational content
Shimmer	Bright and energetic yet gentle voice for friendly interfaces
Verse	Most flexible voice, adapting to various speaking styles and contexts
Marin	Smooth and expressive voice with enhanced naturalness, perfect for dynamic interactions
Cedar	Rich and grounded voice with natural warmth, ideal for conversational AI and customer engagement

Getting Started

Core Concepts

Accounts & Billing

Features

Resources

Reasoning Tiers

Text Models

Realtime Voice Models

Available Voices

​Reasoning Tiers

​Text Models

​Realtime Voice Models

​Available Voices

Reasoning Tiers

Text Models

Realtime Voice Models

Available Voices