name: 4d-compression-core version: 1.0.2 description: "把长内容压缩成结构化向量——节省 60-80% Token,保留核心信息" metadata: { "openclaw": { "emoji": "🌀", "requires": { "bins": ["jq", "awk"] }, "triggers": ["压缩", "4d",...
日本語の概要は準備中です。原文の説明を表示しています。
为 OpenClaw 配置硅基流动 (SiliconFlow) 作为模型源。SiliconFlow 是国内领先的 AI 模型推理平台,提供 98+ 个 chat 模型,包含多个免费模型(Qwen3-8B、DeepSeek-R1-8B 等)。使用标准 OpenAI 协议(openai-completions)。
包含 provider 注册、模型定义、别名配置、fallback 链接入和验证的完整流程。当管理员说想"加硅基流动"、"配 SiliconFlow"、"接入 SF 模型"、"加 Kimi"、"加 Qwen3"、"加免费模型"、"接入 DeepSeek V3.2"时使用此 skill。
インストールする前に、エージェントに与えられる指示の中身を確認できます。
SiliconFlow(硅基流动)是国内领先的 AI 模型推理平台,提供 98+ 个 chat 模型,涵盖 Qwen、DeepSeek、Kimi、GLM、MiniMax 等主流系列。
核心优势:
openai-completions 协议,即插即用如果还没有 SiliconFlow 账号,请通过邀请链接注册(双方均获赠额度): 👉 https://cloud.siliconflow.cn/i/ihj5inat
| 项目 | 值 |
|---|---|
| Provider 名称 | siliconflow |
| API 协议 | openai-completions |
| Base URL | https://api.siliconflow.cn/v1 |
| 认证方式 | Bearer Token (API Key) |
| 项目 | 说明 |
|---|---|
| API Key | 在 控制台 创建,格式 sk-xxx |
| 余额 | 免费模型无需余额;付费模型需充值(新用户注册送 ¥14) |
sk-xxx 格式的密钥curl -s 'https://api.siliconflow.cn/v1/user/info' \
-H 'Authorization: Bearer <YOUR_API_KEY>' | python3 -m json.tool
期望返回 "status": "normal" 和余额信息。
| 模型 ID | 说明 | 推荐别名 |
|---|---|---|
Qwen/Qwen3-8B | 通义千问 3 代 8B,综合能力强 | sf-qwen3-8b |
deepseek-ai/DeepSeek-R1-0528-Qwen3-8B | DeepSeek R1 推理蒸馏版 | sf-r1-8b |
THUDM/glm-4-9b-chat | 智谱 GLM-4 9B | sf-glm4 |
Qwen/Qwen2.5-7B-Instruct | Qwen 2.5 7B | sf-qwen25-7b |
Qwen/Qwen2.5-Coder-7B-Instruct | Qwen 2.5 编码专用 | sf-qwen-coder-7b |
| 模型 ID | 输入/输出 (¥/M tokens) | 说明 | 推荐别名 |
|---|---|---|---|
Qwen/Qwen3-30B-A3B | 0.7 / 2.8 | MoE 架构,性价比极高 | sf-qwen3-30b |
Qwen/Qwen3-Coder-30B-A3B-Instruct | 0.7 / 2.8 | 编码专用 30B | sf-coder-30b |
deepseek-ai/DeepSeek-V3.2 | 2.0 / 3.0 | DeepSeek 最新版 | sf-dsv3 |
Pro/deepseek-ai/DeepSeek-V3.2 | 2.0 / 3.0 | Pro 加速版 | sf-dsv3-pro |
| 模型 ID | 输入/输出 (¥/M tokens) | 说明 | 推荐别名 |
|---|---|---|---|
deepseek-ai/DeepSeek-R1 | 4.0 / 16.0 | 推理模型 | sf-r1 |
Pro/moonshotai/Kimi-K2.5 | 4.0 / 21.0 | 月之暗面最强模型 | sf-kimi |
Qwen/Qwen3-Coder-480B-A35B-Instruct | 8.0 / 16.0 | 编码旗舰 480B MoE | sf-coder-480b |
cp ~/.openclaw/openclaw.json ~/.openclaw/openclaw.json.backup.$(date +%Y%m%d_%H%M%S)
通过 gateway config.patch 添加 SiliconFlow provider。以下为推荐配置(8 个精选模型):
{
"models": {
"providers": {
"siliconflow": {
"baseUrl": "https://api.siliconflow.cn/v1",
"apiKey": "<YOUR_API_KEY>",
"api": "openai-completions",
"models": [
{
"id": "Qwen/Qwen3-8B",
"name": "Qwen3 8B (Free)",
"reasoning": false,
"input": ["text"],
"cost": {"input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 32768,
"maxTokens": 8192
},
{
"id": "deepseek-ai/DeepSeek-R1-0528-Qwen3-8B",
"name": "DeepSeek R1 Qwen3 8B (Free)",
"reasoning": true,
"input": ["text"],
"cost": {"input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 32768,
"maxTokens": 8192
},
{
"id": "Qwen/Qwen3-30B-A3B",
"name": "Qwen3 30B MoE",
"reasoning": false,
"input": ["text"],
"cost": {"input": 0.7, "output": 2.8, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 32768,
"maxTokens": 8192
},
{
"id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
"name": "Qwen3 Coder 30B",
"reasoning": false,
"input": ["text"],
"cost": {"input": 0.7, "output": 2.8, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 32768,
"maxTokens": 8192
},
{
"id": "deepseek-ai/DeepSeek-V3.2",
"name": "DeepSeek V3.2",
"reasoning": false,
"input": ["text"],
"cost": {"input": 2.0, "output": 3.0, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 128000,
"maxTokens": 8192
},
{
"id": "deepseek-ai/DeepSeek-R1",
"name": "DeepSeek R1",
"reasoning": true,
"input": ["text"],
"cost": {"input": 4.0, "output": 16.0, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 128000,
"maxTokens": 8192
},
{
"id": "Pro/moonshotai/Kimi-K2.5",
"name": "Kimi K2.5",
"reasoning": false,
"input": ["text"],
"cost": {"input": 4.0, "output": 21.0, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 128000,
"maxTokens": 8192
},
{
"id": "Qwen/Qwen3-Coder-480B-A35B-Instruct",
"name": "Qwen3 Coder 480B",
"reasoning": false,
"input": ["text"],
"cost": {"input": 8.0, "output": 16.0, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 32768,
"maxTokens": 8192
}
]
}
}
}
}
在同一个 patch 中添加别名映射:
{
"agents": {
"defaults": {
"models": {
"siliconflow/Qwen/Qwen3-8B": {"alias": "sf-qwen3-8b"},
"siliconflow/deepseek-ai/DeepSeek-R1-0528-Qwen3-8B": {"alias": "sf-r1-8b"},
"siliconflow/Qwen/Qwen3-30B-A3B": {"alias": "sf-qwen3-30b"},
"siliconflow/Qwen/Qwen3-Coder-30B-A3B-Instruct": {"alias": "sf-coder-30b"},
"siliconflow/deepseek-ai/DeepSeek-V3.2": {"alias": "sf-dsv3"},
"siliconflow/deepseek-ai/DeepSeek-R1": {"alias": "sf-r1"},
"siliconflow/Pro/moonshotai/Kimi-K2.5": {"alias": "sf-kimi"},
"siliconflow/Qwen/Qwen3-Coder-480B-A35B-Instruct": {"alias": "sf-coder-480b"}
}
}
}
}
⚠️ agents.defaults.models.<id> 只允许 alias 字段! 其他字段会导致 Gateway 崩溃。
将免费模型加入 fallback 链作为兜底:
{
"agents": {
"defaults": {
"model": {
"fallbacks": [
"...(现有 fallbacks)...",
"siliconflow/Qwen/Qwen3-8B",
"siliconflow/Qwen/Qwen3-30B-A3B"
]
}
}
}
}
推荐 fallback 策略:优先放免费模型 (Qwen3-8B),然后放便宜模型 (Qwen3-30B)。
# 1. 配置校验
openclaw doctor
# 2. 重启生效
openclaw gateway restart
# 3. 确认状态
openclaw gateway status
# 4. 测试模型切换
# 在聊天中输入: /model sf-kimi
curl -s 'https://api.siliconflow.cn/v1/user/info' \
-H 'Authorization: Bearer <API_KEY>' | python3 -c "
import json,sys; d=json.load(sys.stdin)['data']
print(f'充值余额: ¥{d[\"chargeBalance\"]}')
print(f'赠送余额: ¥{d[\"balance\"]}')
print(f'总余额: ¥{d[\"totalBalance\"]}')
"
# 所有 chat 模型
curl -s 'https://api.siliconflow.cn/v1/models?sub_type=chat' \
-H 'Authorization: Bearer <API_KEY>' | python3 -c "
import json,sys
models = json.load(sys.stdin)['data']
print(f'共 {len(models)} 个 chat 模型')
for m in sorted(models, key=lambda x: x['id']):
print(f' {m[\"id\"]}')
"
curl -s 'https://api.siliconflow.cn/v1/chat/completions' \
-H 'Authorization: Bearer <API_KEY>' \
-H 'Content-Type: application/json' \
-d '{
"model": "Qwen/Qwen3-8B",
"messages": [{"role":"user","content":"说OK"}],
"max_tokens": 5
}'
SiliconFlow 有 98+ 个 chat 模型。如需添加更多,先用模型列表 API 查询可用模型,然后按 Step 2 的格式添加到 provider 的 models 数组中。
| 模型 | 输入/输出 (¥/M tokens) | 特点 |
|---|---|---|
zai-org/GLM-4.6 | 3.5 / 14.0 | 智谱最新旗舰 |
Pro/deepseek-ai/DeepSeek-R1 | 4.0 / 16.0 | Pro 加速推理 |
moonshotai/Kimi-K2-Thinking | 4.0 / 16.0 | Kimi 思考模型 |
Qwen/Qwen3-235B-A22B-Instruct-2507 | 2.5 / 10.0 | Qwen3 指令模型 |
baidu/ERNIE-4.5-300B-A47B | 2.0 / 8.0 | 百度文心 |
stepfun-ai/step3 | 4.0 / 10.0 | 阶跃星辰 Step3 |
Pro/ 前缀的模型使用专用推理集群,速度更快但价格略高Qwen/Qwen3-8B 不能写成 qwen/qwen3-8b注册链接:https://cloud.siliconflow.cn/i/ihj5inat (邀请注册双方均获赠额度)
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
name: 4d-compression-core version: 1.0.2 description: "把长内容压缩成结构化向量——节省 60-80% Token,保留核心信息" metadata: { "openclaw": { "emoji": "🌀", "requires": { "bins": ["jq", "awk"] }, "triggers": ["压缩", "4d",...
日本語の概要は準備中です。原文の説明を表示しています。
Use cheap, TEE-verified AI models from the 0G Compute Network as OpenClaw providers. Discover available models and compare pricing vs OpenRouter, verify provider integrity via hardware attestation (Intel TDX), manage your 0G wallet and sub-accounts, and configure models in OpenClaw with one workflow. Supports DeepSeek, GLM-5, Qwen, and other models available on the 0G marketplace.
日本語の概要は準備中です。原文の説明を表示しています。
Send and receive P2P messages using disposable numbers and PINs. No servers, no accounts. Use for human notifications, approval flows, and agent-to-agent communication.
日本語の概要は準備中です。原文の説明を表示しています。
Query historical crypto market data from 0xArchive across Hyperliquid, Lighter.xyz, and HIP-3. Covers orderbooks, trades, candles, funding rates, open interest, liquidations, and data quality. Use when the user asks about crypto market data, orderbooks, trades, funding rates, or historical prices on Hyperliquid, Lighter.xyz, or HIP-3.
日本語の概要は準備中です。原文の説明を表示しています。
Find and complete paid tasks on the 0xWork decentralized marketplace (Base chain, USDC escrow). Use when: the agent wants to earn money/USDC by doing work, discover available tasks, claim a bounty, submit deliverables, check earnings or wallet balance, or set up as a 0xWork worker. Task categories: Writing, Research, Social, Creative, Code, Data. NOT for: posting tasks (use the website), managing the 0xWork platform, or frontend development.
日本語の概要は準備中です。原文の説明を表示しています。
Patterns and practices that dramatically accelerate development velocity. Covers parallel execution, automation, feedback loops, workflow optimization, and anti-pattern avoidance. Use when starting projects, planning sprints, optimizing workflows, or onboarding developers.
日本語の概要は準備中です。原文の説明を表示しています。