56 lines
1.7 KiB
YAML
56 lines
1.7 KiB
YAML
# LiteLLM — Proxy unificado de LLMs para o Arcádia Suite
|
|
# Documentação: https://docs.litellm.ai/docs/proxy/configs
|
|
|
|
model_list:
|
|
|
|
# ── OpenAI (quando disponível) ───────────────────────────────────────────────
|
|
- model_name: gpt-4o
|
|
litellm_params:
|
|
model: openai/gpt-4o
|
|
api_key: os.environ/OPENAI_API_KEY
|
|
|
|
- model_name: gpt-4o-mini
|
|
litellm_params:
|
|
model: openai/gpt-4o-mini
|
|
api_key: os.environ/OPENAI_API_KEY
|
|
|
|
# ── Ollama (LLMs locais — soberania) ─────────────────────────────────────────
|
|
- model_name: llama3.3
|
|
litellm_params:
|
|
model: ollama/llama3.3
|
|
api_base: os.environ/OLLAMA_BASE_URL
|
|
|
|
- model_name: qwen2.5-coder
|
|
litellm_params:
|
|
model: ollama/qwen2.5-coder:7b
|
|
api_base: os.environ/OLLAMA_BASE_URL
|
|
|
|
- model_name: deepseek-r1
|
|
litellm_params:
|
|
model: ollama/deepseek-r1:7b
|
|
api_base: os.environ/OLLAMA_BASE_URL
|
|
|
|
# ── Modelo padrão: tenta OpenAI, fallback para Ollama ───────────────────────
|
|
- model_name: arcadia-default
|
|
litellm_params:
|
|
model: openai/gpt-4o-mini
|
|
api_key: os.environ/OPENAI_API_KEY
|
|
model_info:
|
|
fallbacks: ["llama3.3"]
|
|
|
|
router_settings:
|
|
routing_strategy: least-busy
|
|
fallbacks:
|
|
- {"gpt-4o": ["llama3.3"]}
|
|
- {"gpt-4o-mini": ["qwen2.5-coder"]}
|
|
- {"arcadia-default": ["llama3.3"]}
|
|
|
|
litellm_settings:
|
|
drop_params: true
|
|
request_timeout: 120
|
|
set_verbose: false
|
|
|
|
general_settings:
|
|
master_key: os.environ/LITELLM_MASTER_KEY
|
|
database_url: os.environ/DATABASE_URL
|