klbr/khoj
1
0
Fork 0
mirror of https://github.com/khoaliber/khoj.git synced 2026-03-02 13:18:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update model pricing, default models, context and version metadata

Browse Source
This commit is contained in:
Debanjum
2025-06-10 19:28:04 -07:00
parent 753972997f
commit 9b767438e2
2 changed files with 15 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/khoj/processor/conversation/utils.py
View File

@@ -62,14 +62,15 @@ model_to_prompt_size = {
"gpt-4.1": 60000,
"gpt-4.1-mini": 120000,
"gpt-4.1-nano": 120000,
"o1": 20000,
"o3": 30000,
"o1-mini": 60000,
"o3-mini": 60000,
"o4-mini": 60000,
"o1-mini": 90000,
"o1": 30000,
"o3-mini": 90000,
"o3": 60000,
"o3-pro": 30000,
"o4-mini": 90000,
# Google Models
"gemini-2.5-flash-preview-04-17": 120000,
"gemini-2.5-pro-preview-03-25": 60000,
"gemini-2.5-flash-preview-05-20": 120000,
"gemini-2.5-pro-preview-06-05": 60000,
"gemini-2.0-flash": 120000,
"gemini-2.0-flash-lite": 120000,
"gemini-1.5-flash": 120000,

13
src/khoj/utils/constants.py
View File

@@ -17,8 +17,8 @@ default_offline_chat_models = [
"bartowski/gemma-2-2b-it-GGUF",
"bartowski/Qwen2.5-14B-Instruct-GGUF",
]
default_openai_chat_models = ["gpt-4o-mini", "gpt-4.1"]
default_gemini_chat_models = ["gemini-2.0-flash", "gemini-2.5-flash-preview-05-20", "gemini-2.5-pro-preview-05-06"]
default_openai_chat_models = ["gpt-4o-mini", "gpt-4.1", "o3", "o4-mini"]
default_gemini_chat_models = ["gemini-2.0-flash", "gemini-2.5-flash-preview-05-20", "gemini-2.5-pro-preview-06-05"]
default_anthropic_chat_models = ["claude-sonnet-4-0", "claude-3-5-haiku-latest"]
empty_config = {
@@ -41,10 +41,11 @@ model_to_cost: Dict[str, Dict[str, float]] = {
"gpt-4.1": {"input": 2.00, "output": 8.00},
"gpt-4.1-mini": {"input": 0.40, "output": 1.60},
"gpt-4.1-nano": {"input": 0.10, "output": 0.40},
"o1": {"input": 15.0, "output": 60.00},
"o3": {"input": 10.0, "output": 40.00},
"o1-mini": {"input": 3.0, "output": 12.0},
"o1": {"input": 15.0, "output": 60.00},
"o3-mini": {"input": 1.10, "output": 4.40},
"o3": {"input": 2.0, "output": 8.00},
"o3-pro": {"input": 20.0, "output": 80.00},
"o4-mini": {"input": 1.10, "output": 4.40},
# Gemini Pricing: https://ai.google.dev/pricing
"gemini-1.5-flash": {"input": 0.075, "output": 0.30},
@@ -53,8 +54,8 @@ model_to_cost: Dict[str, Dict[str, float]] = {
"gemini-1.5-pro-002": {"input": 1.25, "output": 5.00},
"gemini-2.0-flash": {"input": 0.10, "output": 0.40},
"gemini-2.0-flash-lite": {"input": 0.0075, "output": 0.30},
"gemini-2.5-flash-preview-04-17": {"input": 0.15, "output": 0.60, "thought": 3.50},
"gemini-2.5-pro-preview-03-25": {"input": 1.25, "output": 10.0},
"gemini-2.5-flash-preview-05-20": {"input": 0.15, "output": 0.60, "thought": 3.50},
"gemini-2.5-pro-preview-06-05": {"input": 1.25, "output": 10.0},
# Anthropic Pricing: https://www.anthropic.com/pricing#anthropic-api
"claude-3-5-haiku-20241022": {"input": 1.0, "output": 5.0, "cache_read": 0.08, "cache_write": 1.0},
"claude-3-5-haiku@20241022": {"input": 1.0, "output": 5.0, "cache_read": 0.08, "cache_write": 1.0},