apunkt/SurfSense
1
0
Fork 0
mirror of https://github.com/MODSetter/SurfSense.git synced 2026-05-03 21:02:40 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

feat: unified credits and its cost calculations

Browse source
This commit is contained in:
DESKTOP-RTLN3BA\$punk 2026-05-02 14:34:23 -07:00
parent 451a98936e
commit ae9d36d77f
61 changed files with 5835 additions and 272 deletions
Download patch file Download diff file Expand all files Collapse all files

28
docker/.env.example
View file

@ -159,10 +159,13 @@ STRIPE_PAGE_BUYING_ENABLED=FALSE
# STRIPE_RECONCILIATION_LOOKBACK_MINUTES=10
# STRIPE_RECONCILIATION_BATCH_SIZE=100
# Premium token purchases ($1 per 1M tokens for premium-tier models)
# Premium credit purchases via Stripe ($1 buys 1_000_000 micro-USD of
# credit; premium turns debit the actual per-call provider cost
# reported by LiteLLM, so cheap and expensive models bill proportionally)
# STRIPE_TOKEN_BUYING_ENABLED=FALSE
# STRIPE_PREMIUM_TOKEN_PRICE_ID=price_...
# STRIPE_TOKENS_PER_UNIT=1000000
# STRIPE_CREDIT_MICROS_PER_UNIT=1000000
# DEPRECATED — STRIPE_TOKENS_PER_UNIT=1000000
# ------------------------------------------------------------------------------
# TTS & STT (Text-to-Speech / Speech-to-Text)
@ -315,9 +318,24 @@ STT_SERVICE=local/base
# Pages limit per user for ETL (default: unlimited)
# PAGES_LIMIT=500
# Premium token quota per registered user (default: 5M)
# Only applies to models with billing_tier=premium in global_llm_config.yaml
# PREMIUM_TOKEN_LIMIT=5000000
# Premium credit quota per registered user, in micro-USD (default: $5).
# Premium turns are debited at the actual per-call provider cost reported
# by LiteLLM. Only applies to models with billing_tier=premium.
# PREMIUM_CREDIT_MICROS_LIMIT=5000000
# DEPRECATED — PREMIUM_TOKEN_LIMIT=5000000
# Safety ceiling on per-call premium reservation, in micro-USD ($1.00 default).
# QUOTA_MAX_RESERVE_MICROS=1000000
# Per-image reservation for POST /image-generations, in micro-USD ($0.05 default).
# QUOTA_DEFAULT_IMAGE_RESERVE_MICROS=50000
# Per-podcast reservation for the podcast Celery task ($0.20 default).
# QUOTA_DEFAULT_PODCAST_RESERVE_MICROS=200000
# Per-video-presentation reservation for the video Celery task ($1.00 default).
# Override path bypasses QUOTA_MAX_RESERVE_MICROS clamp — raise with care.
# QUOTA_DEFAULT_VIDEO_PRESENTATION_RESERVE_MICROS=1000000
# No-login (anonymous) mode — public users can chat without an account
# Set TRUE to enable /free pages and anonymous chat API