News from the IT community on September 26th: OpenCode announced yesterday the launch of the second phase of “Operation Cheepseek”. The quota of $60 (Note from the IT community: the current exchange rate is approximately 403.5 RMB) that was originally available for a limited time in OpenCode Go has now been adjusted to be permanently valid.

Previously, OpenCode temporarily increased the usage limit of DeepSeek V4 .1 Flash to four times the original standard, from $15 to $60, and planned to end this promotion on September 20th. Currently, on the Go page, the limit for this model is marked as $60, with the original $15 limit being crossed out.

DeepSeek The newly released V4.1 and Flash adopt a MoE architecture with 552 billion parameters, featuring 8 billion active parameters on the input side and 16 billion on the output side. They also utilize the new Causal Encoder - Decoder architecture. The model supports native multi-modal visual understanding as well.

Compared to the previous generation, the demand for KV Cache of DeepSeek V4 .1 Flash has decreased to one quarter, and the demand for HBM has decreased to one eighth. The demand for SSD has also decreased to one eighth.

DeepSeek previously also announced that OpenCode has been fully integrated as an official partner with V4.1 and Flash. OpenCode Data shows that as of September 25th, this model ranked first in recent platform usage, accounting for about 13% of the observed usage volume of over 2 million times.