OpenCode Makes DeepSeek V4.1 Flash Free with 200K Context Window

On October 15, OpenCode announced that DeepSeek V4.1 Flash is now free to use. The free tier offers a 200K context window, which is shorter than the model's 1M-context version. Calls to the model on the OpenCode platform no longer count toward quota consumption.

The move extends the company's “Operation Cheepseek” initiative. In September 2026, OpenCode temporarily raised the DeepSeek V4.1 Flash usage allowance in the OpenCode Go plan from $15 to $60, then adjusted it to $80 on October 12.

DeepSeek V4.1 Flash launched on September 10, 2026, as a natively multimodal MoE model with 552B main parameters plus 196B Engram conditional memory parameters. Its architecture uses a causal encoder-decoder design: in the 40-layer language backbone, the first 20 layers are causal encoders and the last 20 are decoders, nearly halving prefill computation. On DeepSeek's official platform, the model has no free API tier and is billed through prepaid usage.

Community discussions have focused on generation speed fluctuations. Some users reported peak speeds around 490 tokens/s on the ollama cloud platform, but speeds were unstable and could drop to 50-60 tokens/s around midday. OpenCode Go users also reported TPS falling below 100. Tests showed speed declining as context grows, dropping below 100 TPS beyond 100K context and to about 50 TPS at 200K.