OpenCode offers DeepSeek V4.1 flash for free with 200k context support

OpenCode has announced that DeepSeek V4.1 Flash is now available for free, with a tested support for 200K context window, a reduction from the 1M context version. This move follows the 'Operation Cheepseek' initiative, where OpenCode temporarily increased the usage quota for DeepSeek V4.1 Flash in the OpenCode Go plan from $15 to $60 in September, and later to $80 in October. The free access means calls to the model on the OpenCode platform no longer count against the usage quota. DeepSeek V4.1 Flash, launched on September 10, 2026, is a native multimodal MoE model with 552B main parameters and 196B Engram conditional memory parameters. It uses a causal encoder-decoder design, with 40 layers, the first 20 as an encoder and the next 20 as a decoder, reducing prefill computation by nearly half. Users have reported significant speed fluctuations, with some noting output speeds as high as 490 tokens/s on ollama cloud but dropping to around 50 tokens/s at 200K context. Some users also reported TPS dropping below 100 in the OpenCode Go plan.