OpenCode Makes DeepSeek V4.1 Flash Free to Use, With a 200K Context Window

OpenCode, the open-source AI coding agent, says DeepSeek V4.1 Flash is now free to use on its platform. Calls to the model inside OpenCode no longer draw down plan credits.

The free version has a smaller context window. According to IT Home's tests, it supports 200K tokens, down from the 1 million tokens available in the full version of the model.

The move builds on what OpenCode calls Operation Cheepseek. In September the company temporarily raised the allowance for DeepSeek V4.1 Flash in its OpenCode Go plan from $15 to $60, then lifted it to $80 on October 12. Free access now replaces the allowance for this model.

The change applies only within OpenCode. DeepSeek's own platform still has no free API tier and continues to bill V4.1 Flash on a prepaid, pay-as-you-go basis. DeepSeek released V4.1 Flash on September 10.

Users say speed is uneven. Several developers reported wide swings in output speed. One user said they saw peaks of about 490 tokens per second on the Ollama cloud but drops to 50 or 60 tokens per second from midday into the afternoon. Within the OpenCode Go plan, some users said speed fell below 100 tokens per second. Community tests also suggest speed drops as context grows, with reports of under 100 tokens per second past 100K tokens and about 50 tokens per second near the 200K limit. These are user reports, not official benchmarks.

OpenCode has not said how long the free offer will last. For developers, the trade-off is free access to a capable coding model in exchange for a shorter context window and less predictable speed than the paid route.

SEE ALSO: DeepSeek Ships V4.1 Flash GA With Causal-Encoder-Decoder MoE as V4 Pro Retires