---
title: "OpenAI Shuts Down the Assistants API on 2026-08-26 and Schedules Agent Builder, Evals, and Reusable Prompts for End of November"
url: https://xcube.enlightcorp.com.tw/en/news/openai-assistants-api-shutdown-platform-deprecations-2026
category: "API Platform / Developers"
source_type: official
published: 2026-08-26
updated: 2026-08-26
lang: en
---

# OpenAI Shuts Down the Assistants API on 2026-08-26 and Schedules Agent Builder, Evals, and Reusable Prompts for End of November

Published 2026-08-26 · Updated 2026-08-26 · Official release · Source: [OpenAI API Deprecations](https://developers.openai.com/api/docs/deprecations)

**Key answer:** OpenAI shut down the Assistants API on August 26, 2026, and will retire v1/prompts reusable prompts, the Evals platform, and Agent Builder on November 30, 2026; this narrows the OpenAI API surface to the Responses API and the new Agents API, so enterprise integrations on the old endpoints must migrate.

The OpenAI API Changelog states that 'The Assistants API shut down on August 26, 2026.' In the same period, OpenAI's Deprecations page lists three retirements announced on 2026-06-03: the v1/prompts API and reusable prompt objects, the Evals platform (read-only from 2026-10-31), and Agent Builder, all shutting down on 2026-11-30. OpenAI recommends moving prompt content into application code, points Evals users to Promptfoo, and points Agent Builder users to the Agents SDK or ChatGPT Workspace Agents.

Transcription is consolidating too. On 2026-08-26 OpenAI deprecated whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe, and gpt-4o-transcribe-diarize, with a 2027-02-26 shutdown and gpt-transcribe or gpt-live-transcribe as replacements. On 2026-10-01 it also announced that gpt-5.3-codex, gpt-5.4-nano, and gpt-5.1 will shut down on 2027-04-01, replaced by gpt-6-sol or gpt-6-luna.

Why it matters: the OpenAI API surface is moving from several parallel systems to a Responses API core, with the Agents API taking over hosted agents. For an ecosystem that treats the OpenAI API as a de facto standard, what is being retired is not just models but whole endpoint families and platform tools. Teams that kept prompt management, evaluations, or agent orchestration on OpenAI's platform now have to decide where those assets live.

Caveat: the Deprecations page is the authoritative source for dates, which differ per item and may change; this brief lists the dates shown at the time of writing. whisper-1 and its siblings have a buffer until 2027-02-26, but Evals becomes read-only on 2026-10-31, which is comparatively tight.

The official dates suggest a migration order. Most urgent is Evals, which becomes read-only on 2026-10-31, so any evaluation records worth keeping should be exported before then. Next are v1/prompts, the Evals platform, and Agent Builder, all shutting down on 2026-11-30: prompt content needs to move back into code or an in-house configuration store, and Agent Builder flows need to be rebuilt on the Agents SDK or another orchestration layer. After that come whisper-1 and the gpt-4o transcription models on 2027-02-26, and gpt-5.3-codex, gpt-5.4-nano, and gpt-5.1 on 2027-04-01. The most practical step at the gateway layer is to log which endpoints and model IDs each application actually calls, so affected callers can be identified before each shutdown date.

What to watch: the long-term role of the Chat Completions API. New models such as GPT-6 Sol and GPT-6 Luna still support Chat Completions, but for some models full tool calling is only available through the Responses API. Whether that gap widens will directly shape how compatibility layers are designed.

## X Cube view

Enterprises that wrap models behind an OpenAI-compatible internal interface should inventory applications still calling Assistants, v1/prompts, or whisper-1, bring prompts and eval assets under their own version control, and treat Responses API compatibility — not only chat/completions — as a gateway priority.

Tags: OpenAI, Assistants API, Deprecation, API Migration, Responses API, Agent Builder, Evals, whisper-1, gpt-4o-transcribe