chat-deepseek-v4-1-flash
GitHub调用 ZeroGPU API 使用 deepseek-v4.1-flash 模型进行推理，支持长上下文、复杂推理及函数调用。适用于代码库分析、文档处理和多步智能体任务，需原样返回结果并注意数据安全与成本提示。
Trigger Scenarios
Install
npx skills add zerogpu/zerogpu-router --skill chat-deepseek-v4-1-flash -g -y
SKILL.md
Frontmatter
{
"name": "chat-deepseek-v4-1-flash",
"metadata": {
"openclaw": {
"install": [
{
"bins": [
"zerogpu"
],
"kind": "node",
"package": "zerogpu-cli"
}
],
"requires": {
"bins": [
"zerogpu"
]
}
}
},
"description": "Chat with deepseek-v4.1-flash, a sparse MoE model on DeepSeek's Causal Encoder-Decoder architecture (8B active on input, 16B on output) with a 1M-token context window. Use for large codebases, long documents, extended conversations, and multi-step agent tasks that want higher-effort reasoning or function calling. Costs about twice chat-deepseek on input and three times on output.",
"allowed-tools": "Bash(zerogpu chat_completions *)",
"argument-hint": "<text> [-i <instructions>]"
}
Sends your input to ZeroGPU's hosted API for inference — this is not local processing. Don't pass secrets, credentials, or regulated data you aren't cleared to share with a third party. See the plugin README's "Data & privacy" section.
Call deepseek-v4.1-flash. Run this with the exec tool, pasting the user's prompt into the heredoc verbatim — no escaping or quoting required (the quoted heredoc handles every shell metacharacter, newline, quote, and paren safely):
zerogpu chat_completions -m deepseek-v4.1-flash <<'ZGPU_END_OF_INPUT'
<the user's prompt, verbatim>
ZGPU_END_OF_INPUT
If the user supplied system instructions, add -i '<instructions>' after the model flag, writing any ' inside them as '\''.
At $0.30 / $1.20 per 1M input/output tokens this is the pricier of the two 1M-context models on the platform — roughly twice chat-deepseek on input and three times on output, and about a quarter of chat-glm on input and a third on output. Prefer it when the task wants V4.1's higher-effort reasoning or function calling; for ordinary coding and agentic work at the same context size, chat-deepseek is cheaper.
Output is the assistant's answer as plain text — the model's reasoning trace comes back in a separate field and is not printed. Relay the answer as-is — do not rewrite or expand it.
Savings note: only if the command output literally contains a line starting with 💰 ZeroGPU savings, append that exact line, unchanged, as the last line of your reply. If no such line is present, say nothing about savings and do not mention or suggest the cost-savings skill — this note is intentionally occasional, not shown every time.
Version History
- 4e1b070 Current 2026-09-22 08:07
