Skip to main content
OpenAI

GPT-3.5-turbo pricing, context window and limits

GPT-3.5-turbo costs $0.5 per million input tokens and $1.5 per million output tokens at OpenAI list price. It accepts up to 16,385 tokens of context and returns up to 4,096 tokens.

  • Model id gpt-3.5-turbo
  • Snapshot

What does GPT-3.5-turbo cost?​

List prices in US dollars per million tokens
Input$0.5
Output$1.5
Cached input (read)$0
Cache write—

US dollars per million tokens. A dash means no price is published.

How much can GPT-3.5-turbo read and write?​

Token limits
Context window16,385
Max output4,096

What can GPT-3.5-turbo do?​

Capabilities
Vision (image input)No
Tool callingNo
Structured outputNo
ReasoningNo
Inputtext
Outputtext

What does a request to GPT-3.5-turbo cost?​

Cost of example requests at list price, in US dollars
A typical request: 1,000 tokens in, 500 out$0.0013
A long request: 10,000 tokens in, 1,000 out$0.0065
1,000 typical requests$1.25
1 million tokens in and 1 million out$2

Input read from the provider's prompt cache costs $0 per million tokens instead of $0.5, 100% less. A request that repeats a 1,000-token prompt from the cache costs $0.00075 instead of $0.0013.

Among 49 OpenAI models with a published price, GPT-3.5-turbo is the 12th cheapest by input price. Its context window is the 44th largest of 48. List prices only; discounts, batch prices and prices above 200K tokens are not shown.

How do I call GPT-3.5-turbo through DVARA?​

Configure the openai provider in the DVARA LLM Gateway and send the model id gpt-3.5-turbo. DVARA, the AI governance platform, applies your policies, audit and cost attribution to the call. See LLM cost attribution and budgets.

In DVARA, if you haven't set your own price for a model, its calls are costed at the price in the catalogue that ships with your DVARA version. A price you set in Flightdeck always wins. You can switch catalogue prices off. Then a model with no price you set is counted as costing nothing. Every cost record says which price it used.

What should I check before relying on this data?​

The weekly review marked this entry. A person saw each mark and kept the model listed.

  • The cached-input price is outside the usual 10–50% of the input price.

Which other OpenAI models are there?​

All 53 OpenAI models · Search the full catalogue

Which models from other providers are similar to GPT-3.5-turbo?​

These are the closest in context window and input price, one per provider.

Snapshot 4 October 2026. DVARA reviews every price weekly before it is published. Get the whole catalogue as models.json. Prices are list prices in US dollars per million tokens; check your provider contract before you budget.