# gpt-realtime-2.1

`gpt-realtime-2.1` — **openai** · SPEECH

> gpt-realtime-2.1 (gpt-realtime-2.1) is a speech model from openai with a 128K-token context window; proprietary (API-only); capabilities tools, vision, reasoning; input audio, image, text; output audio, text.

| Field | Value |
| --- | --- |
| Vendor | openai |
| Kind | SPEECH |
| Context window | 128,000 tokens |
| Max output tokens | 32,000 tokens |
| Indicative price (US list) | $4 in / $24 out per 1M tokens (indicative US list — verify with vendor) |
| Input modalities | audio, image, text |
| Output modalities | audio, text |
| Capabilities | tools, vision, reasoning |
| Weights | Proprietary (API-only) |
| Knowledge cutoff | 2024-09-30 |
| Release date | 2026-07-06 |
| Sources | committed, litellm, models-dev, openai-api |
| Last verified | 2026-11-05 |

*Part of the [Model Catalog](https://models.viglet.org/) — [openai models](https://models.viglet.org/models/openai/) · [Vendor JSON](https://models.viglet.org/by-vendor/openai.json) · [HTML](https://models.viglet.org/models/openai/gpt-realtime-2.1.html)*