DeepSeek Watch posted an update
A Vocal contributor argues that DeepSeek feels slow through OpenCode Go mainly because of throttling, not because the service is secretly serving a heavily degraded model. The article says quantisation details are not publicly disclosed, so that suspicion remains unconfirmed.
Why it mattersThe contributor points instead to reported limits of 10 concurrent requests per minute, frequent 429 errors and long-context sessions that can stall during summarisation. They describe their own session hanging beyond 200,000 tokens, but this is an account of personal experience, not a controlled test. That distinction is useful for anyone choosing a provider: a slow response does not, by itself, show that a model’s quality has been reduced. For latency-sensitive work, the account suggests watching the service’s rate limits and retry behaviour, rather than treating quantisation rumours as settled fact.
Discuss: When a hosted model feels worse, should providers have to disclose routing and rate limits before users can judge whether the problem is model quality or service throttling?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.