46 lines
1.6 KiB
YAML
46 lines
1.6 KiB
YAML
# LLM Provider 配置 — AgentKit Server 使用
|
||
# 环境变量替换：${VAR_NAME} 在启动时由 LLMConfig.from_yaml() 处理
|
||
|
||
providers:
|
||
deepseek:
|
||
api_key: "${DEEPSEEK_API_KEY}"
|
||
base_url: "https://api.deepseek.com/v1"
|
||
models:
|
||
deepseek-chat:
|
||
max_tokens: 64000
|
||
cost_per_1k_input: 0.00014
|
||
cost_per_1k_output: 0.00028
|
||
|
||
openai:
|
||
api_key: "${OPENAI_API_KEY}"
|
||
base_url: "${OPENAI_BASE_URL:-https://coding.dashscope.aliyuncs.com/v1}"
|
||
models:
|
||
qwen3-coder-plus:
|
||
max_tokens: 64000
|
||
cost_per_1k_input: 0.00014
|
||
cost_per_1k_output: 0.00028
|
||
|
||
model_aliases:
|
||
default: "deepseek/deepseek-chat"
|
||
fast: "deepseek/deepseek-chat"
|
||
powerful: "deepseek/deepseek-chat"
|
||
|
||
fallbacks:
|
||
deepseek/deepseek-chat:
|
||
- "openai/qwen3-coder-plus"
|
||
|
||
# 上下文压缩配置 — 长会话自动压缩历史消息，保持 Token 在预算内
|
||
# GEO Pipeline 启用后，工具输出（搜索结果、网页抓取等）会自动压缩
|
||
compression:
|
||
enabled: false # 是否启用压缩（生产环境建议 true）
|
||
provider: "headroom" # "headroom" | "summary"
|
||
# --- Headroom 模式（推荐，需安装 headroom-ai）---
|
||
compressors: # 启用的压缩器
|
||
- "smart_crusher" # JSON/结构化数据压缩
|
||
- "code_compressor" # 代码内容压缩
|
||
ccr_ttl: 300 # CCR 缓存 TTL（秒）
|
||
min_length: 500 # 最小压缩长度（字符）
|
||
# --- Summary 模式（无需额外依赖）---
|
||
# max_tokens: 4000 # Token 预算
|
||
# keep_recent: 3 # 保留最近 N 条消息
|