Top news
Investing in the public benefit OfficialAnthropic NewsOct 18Popular Musk Pledges SpaceX Shares to ‘Every Kid in America’ WebBloomberg TechnologyOct 19Popular Websites are closing their doors to AI agents WebThe Verge AIOct 17 This is the new ChatGPT Voice, powered by GPT Live OfficialOpenAI on YouTubeOct 18 How AI giants will reshape global philanthropy WebAxios AIOct 19 Gemini 4 Argon Really Surprised Me WebTwo Minute PapersOct 19 Google Gemini 4 agents just beat Minecraft with no human intervention WebEngadget AIOct 19 Who could it be? 🧐 OfficialGoogle on YouTubeOct 18 Anthropic gives outside AI watchdog “employee-like” access WebArs Technica AIOct 18 What are self-replicating ‘AI worms’, and why is Anthropic worried about them? WebThe Guardian AIOct 18
GlossaryProductsNew to AI

Streaming

Also written: Continuous streams, Streaming infrastructure

Streaming sends data continuously for immediate use, like live video, instead of waiting for the whole file.

In plain terms

Streaming means sending data in a continuous flow. Instead of waiting for a whole file to download, you can start using it as it arrives. Think about watching a video online. The video plays while more of it is still being sent to your device.

In AI, this often means getting parts of a response from a model as it's being generated. For example, when you type a prompt into a chatbot, you might see the words appear one by one, rather than waiting for the entire answer to be ready.

The expert version

Streaming refers to the continuous transmission and processing of data in real-time, enabling immediate consumption or action without requiring the complete dataset. This contrasts with batch processing, where data is collected and processed in discrete chunks. In the context of large language models, streaming allows tokens to be sent back to the user as they are generated, improving perceived latency and enabling interactive applications.

The primary benefit is reduced latency, making interactions feel more responsive. A tradeoff can be increased complexity in managing the data flow and handling partial results. For models, streaming is typically implemented by returning an iterator or generator that yields tokens or chunks of text as they become available.

Why it matters now

Frontier AI labs are increasingly using streaming to provide real-time user experiences. This includes live translation, speech-to-text, and generating AI responses as they are created, rather than waiting for the full output. This approach improves responsiveness and allows for more dynamic interactions with AI models.

Where it shows up 7 items, newest first
How each company uses it
Google Gemini
Google Gemini API is adding streaming capabilities for video outputs and enabling real-time voice and video interaction with its models.
Microsoft Copilot
Microsoft Copilot is exploring the use of local AI models, which can benefit from streaming infrastructure for continuous data flow.
OpenAI ChatGPT
OpenAI offers models like GPT-Realtime-Translate and GPT Live Transcribe that provide live, streaming speech-to-speech and speech-to-text translation.
xAI Grok
xAI Grok's Responses API uses streaming to maintain reasoning history for better multi-turn AI conversations.
Related terms

LatencyTokensReal-timeAPIsGenerative AILLM

Often confused with
Batch Processing
Batch processing handles data in large chunks all at once, whi
Buffering
Buffering is a technique used in streaming to temporarily store data
Downloading
Downloading typically means getting an entire file before you can u
Origin

The concept of streaming originated in the early days of the internet for media playback, allowing users to consume audio and video content without downloading the entire file first. It has since been adopted across many computing fields for efficient data handling. Source ↗

Try this about 5 minutes

Try using a chatbot like ChatGPT or Google Gemini. When you ask a question, notice how the answer often appears word by word, rather than all at once. This is an example of streaming output from the AI model.

Try it: Playbook entries that use thisPlaybook →

The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.

Questions people ask

Why do AI models stream responses?
AI models stream responses to make interactions feel faster and more natural. Users see results appearing immediately, improving the experience.
Is streaming only for text?
No, streaming can be used for various data types, including audio and video, as seen in real-time translation or video processing applications.

First seen on Subvolts Sep 26, 2019. Updated Oct 18, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.