With the rapid spread of artificial intelligence technology, free and open source once became a core engine driving the industry's prosperity. However, when a massive user base collides with limited computing resources, real-world friction also surfaces. Recently, DeepSeek, a leading domestic AI player, unexpectedly triggered a wave of negative reviews in app stores by throttling certain ultra-long-context features, and even extreme complaint incidents occurred, reflecting ecosystem pain points behind the current free-access model of large models that cannot be ignored.
Long Conversations and Role-Playing Drive Compute Explosion: Throttling Adjustment Sparks User Dissatisfaction
Previously, DeepSeek attracted an influx of massive users with its outstanding model capabilities and long-term free strategy. In actual use, many users widely applied it to high-intensity, ultra-long-context interactive scenarios such as long novel writing and deep role-playing.
Because such high-frequency and massive interactions frequently trigger enormous Token consumption, they place tremendous load pressure on the platform's underlying computing resources. To maintain the overall stability and sustainable operation of the service, the platform recently began imposing necessary throttling and usage quota limits on some long-running conversation and ultra-long-context features. Although officials emphasized that everyday ordinary queries and standard usage are completely unaffected, this adjustment still triggered strong backlash from some high-frequency heavy users.
Extreme Emotions Expose Ecosystem Weaknesses: "Free" Should Not Equal "Taken for Granted"
In this storm caused by throttling, a large number of negative reviews flooded some app stores, even mixed with extreme individual cases. Some extreme negative voices complained to government departments and even threatened to take extreme actions if the issue was not resolved, ultimately prompting local police and emergency responders to intervene.
This incident sparked widespread lament and heated discussion on social platforms. Many industry insiders and technology observers, citing the ancient saying "A small favor earns gratitude, a large favor breeds resentment," pointed out that the team represented by Liang Wenfeng has long adhered to an open-source and free inclusive path, with the original intention of letting more people enjoy the dividends brought by artificial intelligence. But some users treat free high-quality service as a matter of course, and once reasonable adjustments at the boundaries of technology and resources are triggered, they go to extremes. This also sounded a warning bell for the entire large-model industry: while advancing technological equality and free inclusive access, how to guide the public to form rational expectations about technology and build a healthy and sustainable supply-demand relationship has become a new issue facing all AI practitioners.
Join Now