GPT-3.5 Turbo
OpenAI's legacy GPT-3.5 Turbo: the original model behind ChatGPT. Features message-based ChatML input, fine-tuning, and stable alias versions.
Model Overview
Capabilities, design details, and architectural traits
GPT-3.5 Turbo - The Same Model Behind ChatGPT, Opened Up Through the API
GPT-3.5 Turbo is documented by OpenAI as the same model used inside the ChatGPT product, made available to developers through a dedicated API endpoint. It introduced a message based way of talking to a GPT model, rather than a single block of plain text.
| Trait | Detail |
|---|---|
| Defining identity | The same model used in the ChatGPT product, released through a dedicated API endpoint rather than only through the ChatGPT interface. |
| Structured message format | Consumes a sequence of role tagged messages instead of unstructured text, rendered internally through a format OpenAI calls ChatML. |
| Stable alias versioning | Calling the unversioned gpt-3.5-turbo name returns OpenAI's recommended stable snapshot, while developers can pin a specific dated version, such as gpt-3.5-turbo-0301, for consistent behavior. |
| Early fine-tuning access | Documented as the first of OpenAI's current chat models opened up for supervised fine-tuning, with fine-tuning for GPT 4 following later. |
Originally Launched as the ChatGPT API
When OpenAI first released this model to developers, it was marketed as the ChatGPT API. OpenAI later updated its own documentation to clarify that earlier mentions of the ChatGPT API all refer to the GPT-3.5 Turbo API, retiring the original name.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | OpenAI | Anthropic | Anthropic |
| Release Date | November 30, 2022 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | Sep 2021 | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 16K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $0.50 Best Input Pricing | $4 | $2 |
| Output Pricing | $1.50 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | text | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 5.5 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | 10.7 | - | - |
| Agentic Index | - | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from OpenAI
Other models by OpenAI
Top AI Models
Leading alternatives by intelligence score
