Deep Research doesn’t use Deep Research anymore, it redirects to Agent Mode and it sucks

Understanding the Shift in AI Research Tools: From Deep Research to Agent Mode

In recent developments within AI-powered research tools, users have reported significant changes that impact their workflows and expectations. Notably, the formerly available feature “Deep Research” appears to have been deprecated or replaced by an “Agent Mode,” which, according to many accounts, does not meet the same standards of functionality or performance.

Background and User Experience

Previously, users leveraging services like ChatGPT Plus enjoyed the ability to perform extensive, long-form research seamlessly. One popular use case involved generating comprehensive, narrative-driven content—such as fanfiction—by providing detailed prompts. These prompts encompassed specific parameters like story length, emotional tone, character inclusion, and plot details. The AI would then craft a coherent, engaging story in a single session, often reflecting on its output to ensure quality and alignment with the user’s specifications. This process enabled on-demand storytelling that resonated well with individual preferences.

However, recent observations indicate a decline in the efficacy of this feature. Users report that prompts which previously yielded high-quality narratives now produce subpar or nonsensical results. In some cases, the AI resorts to rudimentary code generation—such as assembling repetitive templates or scripts—instead of crafting the intended stories. This substantial drop in output quality has been confirmed by multiple users, including those with active subscriptions and remaining usage credits.

Changes in Usage and Functionality

A notable issue is the apparent misalignment between the labeled usage quotas and actual consumption. Users have observed that their “Deep Research” credits remain unchanged despite performing tasks that should logically consume them. Instead, these activities seem to deplete “Agent Mode” credits, even when the latter was not intentionally activated. For example:

• Selecting “Deep Research” prompts no longer trigger the expected usage deductions.
• Initiating what should be a research-oriented request consumes “Agent Mode” credits.
• The interface no longer prompts for clarifications or displays progress indicators associated with “Deep Research.”

Implications for Users

These modifications have caused considerable frustration within the user community. Long-standing use cases—particularly those involving lengthy, research-intensive writing—are now hindered by reduced output quality and unclear operational changes. Users engaged in creative projects like fanfiction generation find their tools’ capabilities significantly diminished, leading to disappointment and concern over the platform’s future utility.

Looking for Alternatives

Given these shifts, users are actively seeking alternative AI-powered research and content generation tools that can handle complex, one-shot prompts with reliable results