[app]
# app settings are global available like `r2r_config.agent.app`
# project_name = "r2r_default" # optional, can also set with `R2R_PROJECT_NAME` env var
default_max_documents_per_user = 10_000
default_max_chunks_per_user = 10_000_000
default_max_collections_per_user = 5_000

# Set the default max upload size to 200 GB for local testing
default_max_upload_size = 214748364800

# LLM used for internal operations, like deriving conversation names
fast_llm = "openai/gpt-4o-mini"

# LLM used for user-facing output, like RAG replies
quality_llm = "openai/gpt-4o"

# LLM used for ingesting visual inputs
vlm = "openai/gpt-4o"

# LLM used for transcription
audio_lm = "openai/whisper-1"

# Reasoning model, used for `research` agent
reasoning_llm = "openai/o3-mini"
# Planning model, used for `research` agent
planning_llm = "anthropic/claude-3-7-sonnet-20250219"


[agent]
rag_agent_static_prompt = "static_rag_agent"
rag_agent_dynamic_prompt = "dynamic_rag_agent"
# The following tools are available to the `rag` agent
rag_tools = ["search_file_descriptions", "search_file_knowledge", "get_file_content"] # can add  "web_search" | "web_scrape"
# The following tools are available to the `research` agent
research_tools = ["rag", "reasoning", "critique", "python_executor"]

[auth]
provider = "r2r"
access_token_lifetime_in_minutes = 60000
refresh_token_lifetime_in_days = 7
require_authentication = false
require_email_verification = false
default_admin_email = "admin@example.com"
default_admin_password = "change_me_immediately"

[completion]
provider = "r2r"
concurrent_request_limit = 64

  [completion.generation_config]
  temperature = 0.1
  top_p = 1
  max_tokens_to_sample = 4_096
  stream = false
  add_generation_kwargs = { }

[crypto]
provider = "bcrypt"

[database]
default_collection_name = "Default"
default_collection_description = "Your default collection."
collection_summary_prompt = "collection_summary"

  [database.graph_creation_settings]
    graph_entity_description_prompt = "graph_entity_description"
    graph_extraction_prompt = "graph_extraction"
    entity_types = [] # if empty, all entities are extracted
    relation_types = [] # if empty, all relations are extracted
    automatic_deduplication = true # enable automatic deduplication of entities

  [database.graph_enrichment_settings]
    graph_communities_prompt = "graph_communities"

[embedding]
provider = "litellm"
# For basic applications, use `openai/text-embedding-3-small` with `base_dimension = 512`
# For advanced applications, use `openai/text-embedding-3-large` with `base_dimension = 3072` and binary quantization
base_model = "openai/text-embedding-3-small"
base_dimension = 512
# rerank_model = "huggingface/mixedbread-ai/mxbai-rerank-large-v1" # reranking model
batch_size = 128
add_title_as_prefix = false
concurrent_request_limit = 256
initial_backoff = 1.0
quantization_settings = { quantization_type = "FP32" }

[completion_embedding]
# Generally this should be the same as the embedding config, but advanced users may want to run with a different provider to reduce latency
provider = "litellm"
base_model = "openai/text-embedding-3-small"
base_dimension = 512
batch_size = 128
add_title_as_prefix = false
concurrent_request_limit = 256

[ingestion]
provider = "r2r"
chunking_strategy = "recursive"
chunk_size = 1_024
chunk_overlap = 512
excluded_parsers = ["mp4"]
automatic_extraction = true # enable automatic extraction of entities and relations

  [ingestion.chunk_enrichment_settings]
    chunk_enrichment_prompt = "chunk_enrichment"
    enable_chunk_enrichment = false # disabled by default
    n_chunks = 2 # the number of chunks (both preceding and succeeding) to use in enrichment

  [ingestion.extra_parsers]
    pdf = "zerox"

[logging]
provider = "r2r"
log_table = "logs"
log_info_table = "log_info"

[orchestration]
provider = "simple"

[email]
provider = "console_mock" # `smtp` | `sendgrid` supported