DeepSeek V4.1 Flash in Claude Code and OpenCode: setup, test runs, and what it costs on four platforms

DeepSeek's newest V4.1 Flash snapshot is now accessible via DeepSeek's API, identified by the model name "deepseek-flash." This report explores the performance of this model when used in coding agents, as well as the associated costs on four different platforms.

To test the model's suitability for coding agents, a small task was executed three times on three different platforms: Claude Code, OpenCode, and XHuoAPI. Claude Code took between 16 and 27 seconds per run, while OpenCode completed the task in 9 to 11 seconds. Claude Code required approximately 32k to 55k input tokens per run, with the input cost having a greater impact on pricing than the output cost for agent usage.

The cost of using DeepSeek's V4.1 Flash model varies depending on the platform:

1. DeepSeek's own API: $0.30 / $1.20 per 1 million input/output tokens during peak hours (01:00–04:00 and 06:00–10:00 UTC, Monday to Friday) and $0.15 / $0.60 off-peak.

2. DeepInfra: $0.14 / $0.42 per 1 million input/output tokens, with a promotional rate of $0.20 / $0.60 without the promotion.

3. OpenRouter: $0.30 / $1.20 per 1 million input/output tokens on DeepSeek's endpoint, but the cheapest provider is $0.07 / $0.15, mostly using fp4/fp8 quantization.

4. XHuoAPI: $0.24 / $0.96 per 1 million input/output tokens in the standard tier, with cheaper options like Saver ($0.15 / $0.60) and Spot ($0.09 / $0.36) tiers available, although they come with reduced stability.

Considering cost-efficiency, the best options are OpenRouter's low-cost providers or XHuoAPI's Spot tier, followed by DeepInfra's promotional pricing and DeepSeek's off-peak hours. For first-party usage, DeepSeek's own API is recommended. OpenRouter and XHuoAPI offer unified access to multiple models, including DeepSeek's V4.1 Flash model.

Both platforms allow users to switch between models such as Qwen, GLM, Kimi, GPT, and Claude using a single API key. On XHuoAPI, the Saver tier corresponds to DeepSeek's off-peak pricing at any given time, enabling consistent cost savings.