GPT-3.5 Turbo Instruct
GPT-3.5 Turbo Instruct is an instruction-tuned completion model designed for the legacy Completions endpoint, offering a direct prompt-in, text-out format suited to few-shot tasks, templated generation, and workflows that predate the chat message structure.
import { streamText } from 'ai'
const result = streamText({ model: 'openai/gpt-3.5-turbo-instruct', prompt: 'Why is the sky blue?'})Playground
Try out GPT-3.5 Turbo Instruct by OpenAI. Usage is billed to your team at API rates. Free users (those who haven't made a payment) get $5 of credits every 30 days.
Providers
Route requests across multiple providers. Copy a provider slug to set your preference. Visit the docs for more info. Using a provider means you agree to their terms, listed under Legal.
P50 throughput on live AI Gateway traffic, in tokens per second (TPS). Visit the docs for more info.
P50 time to first token (TTFT) on live AI Gateway traffic, in milliseconds. View the docs for more info.
Direct request success rate on AI Gateway and per-provider. Visit the docs for more info.
More models by OpenAI
| Model |
|---|
| 1M | 1.2s | 74tps | $5.00/M | $30.00/M | Read: $0.5/M Write: — | $10.00/K + input costs | — |   | 04/24/2026 | ||||
| 400K | 1.3s | 176tps | $0.75/M | $4.50/M | Read:$0.07/M Write:— | $10.00/K + input costs | — | | 03/17/2026 | ||||
| 400K | 0.6s | 82tps | $0.20/M | $1.25/M | Read:$0.02/M Write:— | $10.00/K + input costs | — | | 03/17/2026 | ||||
| 1.1M | 0.5s | 71tps | $2.50/M | $15.00/M | Read: $0.25/M Write: — | $10.00/K + input costs | — | | 03/05/2026 | ||||
| 128K | 0.8s | 102tps | $1.25/M | $10.00/M | Read:$0.13/M Write:— | $10.00/K + input costs | — | | 11/12/2025 | ||||
| 400K | 3.2s | 163tps | $0.25/M | $2.00/M | Read:$0.03/M Write:— | $14/K + input costs | — | | 08/07/2025 |
About GPT-3.5 Turbo Instruct
GPT-3.5 Turbo Instruct launched on September 28, 2023 as part of OpenAI's update that also brought fine-tuning support to the GPT-3.5 Turbo family. While the base GPT-3.5 Turbo targets the Chat Completions API, this variant works with the legacy Completions endpoint, the same interface GPT-3 era models used. That makes it the natural successor for workflows built on that format.
The model follows the instructed-completion paradigm: you write a free-form prompt and the model continues or responds directly, without requiring the message-role structure the chat endpoint demands. This suits few-shot pattern completion, structured data extraction from templated prompts, text transformation tasks, and any integration where a prompt-in, completion-out contract is preferable.
OpenAI positioned the Instruct variant as a bridge for developers who wanted Turbo-family efficiency and pricing without abandoning the Completions interface. For new projects, chat-format models generally offer more capability per token. But for existing pipelines or use cases where single-turn completion is semantically cleaner, GPT-3.5 Turbo Instruct fills a clear slot.
What To Consider When Choosing a Provider
- Configuration: If you're migrating legacy Completions-based integrations or maintaining backwards-compatible pipelines, note that this model's prompt format differs from chat models. Preserving your existing prompt templates avoids rework.
- Zero Data Retention: AI Gateway does not currently support Zero Data Retention for this model. See the documentation for models that support ZDR.
- Authentication: AI Gateway authenticates requests using an API key or OIDC token. You do not need to manage provider credentials directly.
When to Use GPT-3.5 Turbo Instruct
Best For
- Legacy completions integrations: Existing Completions-endpoint deployments where migrating to the chat message format isn't yet feasible
- Few-shot text transformation: Prompt templates with examples baked in for pattern completion
- Code completion: Inline suggestion features that use a fill-in-the-middle or suffix-based prompt style
- Structured extraction: Well-crafted prompts that reliably produce a specific output format
- Legacy chatbot pipelines: Automation pipelines built before the Chat Completions API became standard
Consider Alternatives When
- New projects: Starting a new project where GPT-3.5 Turbo or GPT-4o mini will generally offer better multi-turn coherence
- Chat-only features: The task requires vision input, function calling, or JSON mode, which are only available on chat-format models
- Deeper reasoning: You need the extended context or reasoning depth of the GPT-4 family
Conclusion
GPT-3.5 Turbo Instruct occupies a specific but important position: it is the only GPT-3.5 Turbo variant designed for the Completions endpoint, making it the right choice when a completion-style prompt contract is required or preferred. Through AI Gateway, teams can continue using this model alongside newer chat-format models under a single authentication layer.
Frequently Asked Questions
What is the key difference between GPT-3.5 Turbo and GPT-3.5 Turbo Instruct?
GPT-3.5 Turbo uses the Chat Completions endpoint with a messages array. GPT-3.5 Turbo Instruct uses the legacy Completions endpoint with a single prompt string, a fundamental structural difference that affects how you construct requests.
Does GPT-3.5 Turbo Instruct support function calling or JSON mode?
No. Function calling and JSON mode are features of the Chat Completions API. GPT-3.5 Turbo Instruct targets the Completions endpoint and doesn't support these capabilities.
When would I choose the Completions format over Chat Completions?
Use it when your prompt structure works best as a single string with few-shot examples inline, when you're maintaining an existing integration, or when a direct prompt-response contract is semantically simpler than a roles-based message list.
Is GPT-3.5 Turbo Instruct suitable for code completion features?
Yes. Single-turn code completion and fill-in-the-middle tasks map naturally to the Completions format, and the model's instruction tuning makes it responsive to explicit directives within the prompt.
How do I access GPT-3.5 Turbo Instruct through AI Gateway?
Authenticate with an AI Gateway API key or OIDC token and route requests to the AI Gateway endpoint specifying this model's slug. No direct OpenAI credentials are required in your application.
Can GPT-3.5 Turbo Instruct be used for multi-turn conversations?
Technically you can simulate turns by concatenating prior exchanges into a single prompt string, but the chat-format models handle multi-turn context more naturally and efficiently.
What are typical latency characteristics?
This page shows live throughput and time-to-first-token metrics measured across real AI Gateway traffic.