smartproxy/smartproxy.yaml.example at main · balyakin/smartproxy · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
version: 1

listen: "127.0.0.1:4100"
admin_listen: "127.0.0.1:4101"

server:
  shutdown_timeout: 30s
  read_timeout: 120s
  write_timeout: 120s
  read_header_timeout: 15s
  max_body_bytes: 20971520
  max_concurrent_upstream: 0

database:
  path: "~/.smartproxy/smartproxy.db"
  retention_days: 90

providers:
  - name: deepseek
    type: openai
    base_url: "https://api.deepseek.com/v1"
    api_key_env: DEEPSEEK_API_KEY
    max_idle_conns_per_host: 10
    max_conns_per_host: 0

  - name: openai
    type: openai
    base_url: "https://api.openai.com/v1"
    api_key_env: OPENAI_API_KEY

  - name: anthropic
    type: anthropic
    base_url: "https://api.anthropic.com"
    api_key_env: ANTHROPIC_API_KEY
    anthropic_version: "2023-06-01"

targets:
  - name: deepseek-flash
    enabled: true
    provider: deepseek
    model: "deepseek-chat"
    timeout: 120s
    max_retries: 2
    max_context: 64000
    required_capabilities: [chat, stream, tools]

  - name: gpt-mini
    enabled: true
    provider: openai
    model: "gpt-4o-mini"
    timeout: 120s
    max_retries: 2
    max_context: 128000
    required_capabilities: [chat, stream, tools, vision]

  - name: claude-sonnet
    enabled: true
    provider: anthropic
    model: "claude-sonnet-4-20250514"
    timeout: 120s
    max_retries: 2
    max_context: 200000
    required_capabilities: [chat, stream, tools, vision]

model_aliases:
  claude-sonnet: "claude-sonnet-4-20250514"
  cheap: "deepseek-chat"

routes:
  - name: vision-to-claude
    endpoints: ["/v1/chat/completions"]
    strategy: priority
    match:
      has_images: true
    targets: [claude-sonnet, gpt-mini]

  - name: cheap-text-default
    endpoints: ["/v1/chat/completions"]
    strategy: priority
    cache:
      enabled: true
      ttl: 30m
    targets: [deepseek-flash, gpt-mini, claude-sonnet]

failover:
  on_status: [408, 409, 425, 429, 500, 502, 503, 504]
  on_timeout: true
  backoff_base: 250ms
  backoff_max: 5s
  respect_retry_after: true
  circuit_breaker:
    error_threshold: 3
    success_threshold: 2
    min_requests: 3
    window: 60s
    sleep_window: 30s
    jitter: 0.2
    half_open_max_requests: 1

streaming:
  mid_stream_policy: close
  client_notice_text: "\n\n[smartproxy: upstream stream interrupted]"
  notice_for_tool_calls: false

cache:
  enabled: false
  ttl: 1h
  max_size_mb: 500
  max_entries: 100000
  gzip: true
  encrypt: false
  encryption_key_env: SMARTPROXY_CACHE_KEY

budget:
  daily_limit_usd: 10.0
  monthly_limit_usd: 200.0
  session_limit_usd: 0
  alert_at_percent: 80
  action: stop

pricing:
  unknown_model:
    mode: estimate
    input_per_million: 2.0
    output_per_million: 8.0
  overrides:
    my-custom-model:
      input_per_million: 1.0
      output_per_million: 5.0

observability:
  level: info
  log_format: text
  response_headers: true
  log_prompts: false
  log_responses: false
  redact_secrets: true
  metrics:
    enabled: false

security:
  allow_client_authorization_passthrough: false
  admin_bind_required_loopback: true
  tls:
    enabled: false
    cert_file: ""
    key_file: ""
  admin_auth:
    api_key_env: SMARTPROXY_ADMIN_API_KEY
  client_auth:
    mode: loopback_optional
    api_key_env: SMARTPROXY_API_KEY
    require_for_non_loopback: true