OpenAI adds 'Read Aloud' voiceover to ChatGPT

Everyone was waiting for OpenAI to drop GPT-5—or at least GPT-4.5—to counter Anthropic's brand-new Claude 3. Instead, OpenAI quietly rolled out a practical quality-of-life tweak: ChatGPT can now read its answers out loud.

The new feature is officially called Read Aloud. It is available right now on the ChatGPT mobile apps for iOS and Android, with rollout underway on the web version as well.

A long press to start listening

The mechanic is dead simple. On your phone, touch and hold any response from ChatGPT, then tap Read Aloud from the pop-up menu.

On the web, OpenAI added a read-aloud button right below the generated text output. One click starts the narration.

An OpenAI spokesperson told VentureBeat that the feature is meant for hands-free moments, users on the go, or situations where listening to text helps you absorb information better.

Tech outlet The Decoder pointed out that the synthetic voice does not sound quite as polished or realistic as ElevenLabs. That tradeoff makes sense, given the massive compute costs of running voice generation at ChatGPT scale.

Polishing the app while rivals chase benchmarks

The timing is impossible to ignore. Anthropic just crowned Claude 3 as the new leader over GPT-4, Elon Musk is suing OpenAI, and Google is taking heat over Gemini guardrails.

Many expected OpenAI to hit back with raw horsepower. Instead, they added an everyday audio layer to their flagship chatbot.

ChatGPT already had real-time two-way voice conversations on mobile, but that works like a phone call. Read Aloud is built specifically for digesting long written answers without staring at a screen.

Why it matters

We think this update proves OpenAI is shifting real attention to everyday usability rather than just fighting benchmark wars. For regular users, having ChatGPT dictate a long draft, summary, or itinerary while you cook or commute is an immediate quality-of-life win.

It also nudges competitors to look beyond raw LLM intelligence. When model capabilities start to converge, interface friction becomes the real differentiator for retaining casual users.

OpenAI's call to use a slightly more basic synthetic voice looks like a smart economic choice. Keeping latency low and server bills manageable matters far more than studio-grade voice acting for an in-line feature. The catch is that utility features can only buy so much breathing room if rivals keep chipping away at GPT-4's performance crown.