RomanNum3ral
/
cai
mirror of https://github.com/aliasrobotics/cai.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Noticed a difference in the model-list in completer.py and model.py (#230) 

* fix

* reverted parts of the previous changes as they led to a duplication of model-show
* now /model shows the predefined models (as before)
* the numbering in the cached-models is consistent with model-show and the completer

---------

Co-authored-by: christian.pritzl <christian.pritzl@iu.org>
This commit is contained in:
Chris 2025-07-24 09:24:21 +02:00 committed by GitHub
parent 83ab74f585
commit b8dd50dd27
3 changed files with 196 additions and 226 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

94
src/cai/repl/commands/compact.py
View File

@ -7,9 +7,17 @@ from typing import List, Optional
import os
import datetime
from rich.console import Console
from rich.table import Table
from rich.panel import Panel
from cai.repl.commands.base import Command, register_command
from cai.sdk.agents.models.openai_chatcompletions import get_current_active_model
from cai.util import COST_TRACKER
from cai.repl.commands.model import (
ModelCommand,
get_predefined_model_categories,
get_all_predefined_models
)
console = Console()
@ -82,11 +90,6 @@ class CompactCommand(Command):
def handle_model(self, args: Optional[List[str]] = None) -> bool:
"""Set model for compaction."""
from cai.repl.commands.model import ModelCommand
from rich.table import Table
from rich.panel import Panel
from cai.util import COST_TRACKER
if not args:
# Display current model
console.print(
@ -97,85 +100,8 @@ class CompactCommand(Command):
)
)
# Create model command instance to reuse its model data
model_cmd = ModelCommand()
# Define model categories (same as in model.py)
MODEL_CATEGORIES = {
"Alias": [
{
"name": "alias0",
"description": "Best model for Cybersecurity AI tasks"
}
],
"Anthropic Claude": [
{
"name": "claude-sonnet-4-20250514",
"description": "Excellent balance of performance and efficiency"
},
{
"name": "claude-3-7-sonnet-20250219",
"description": "Excellent model for complex reasoning and creative tasks"
},
{
"name": "claude-3-5-sonnet-20240620",
"description": "Excellent balance of performance and efficiency"
},
{
"name": "claude-3-5-haiku-20240307",
"description": "Fast and efficient model"
},
],
"OpenAI": [
{
"name": "o3-mini",
"description": "Latest mini model in the O-series"
},
{
"name": "gpt-4o",
"description": "Latest GPT-4 model with improved capabilities"
},
],
"DeepSeek": [
{
"name": "deepseek-v3",
"description": "DeepSeek's latest general-purpose model"
},
{
"name": "deepseek-r1",
"description": "DeepSeek's specialized reasoning model"
}
]
}
# Create a flat list of all models
ALL_MODELS = []
for category, models in MODEL_CATEGORIES.items():
for model in models:
# Get pricing info
input_cost_per_token, output_cost_per_token = COST_TRACKER.get_model_pricing(model["name"])
# Convert to dollars per million tokens
input_cost_per_million = None
output_cost_per_million = None
if input_cost_per_token is not None and input_cost_per_token > 0:
input_cost_per_million = input_cost_per_token * 1000000
if output_cost_per_token is not None and output_cost_per_token > 0:
output_cost_per_million = output_cost_per_token * 1000000
ALL_MODELS.append({
"name": model["name"],
"provider": (
"Anthropic" if "claude" in model["name"]
else "DeepSeek" if "deepseek" in model["name"]
else "OpenAI"
),
"category": category,
"description": model["description"],
"input_cost": input_cost_per_million,
"output_cost": output_cost_per_million
})
# Get all predefined models using the shared function
ALL_MODELS = get_all_predefined_models()
# Show available models in a table
model_table = Table(

71
src/cai/repl/commands/completer.py
View File

@ -30,6 +30,7 @@ from cai.repl.commands.base import (
COMMANDS,
COMMAND_ALIASES
)
from cai.repl.commands.model import get_predefined_model_names
console = Console()
@ -82,12 +83,12 @@ class FuzzyCommandCompleter(Completer):
def _background_fetch_models(self):
"""Fetch models in background to avoid blocking the UI."""
try:
self.fetch_ollama_models()
self.fetch_all_models()
except Exception: # pylint: disable=broad-except
pass
def fetch_ollama_models(self): # pylint: disable=too-many-branches,too-many-statements,inconsistent-return-statements,line-too-long # noqa: E501
"""Fetch available models from Ollama if it's running."""
def fetch_all_models(self): # pylint: disable=too-many-branches,too-many-statements,inconsistent-return-statements,line-too-long # noqa: E501
"""Fetch all available models (predefined + LiteLLM + Ollama) to match /model command."""
# Only fetch every 60 seconds to avoid excessive API calls
now = datetime.datetime.now()
@ -97,8 +98,28 @@ class FuzzyCommandCompleter(Completer):
return
self._last_model_fetch = now
ollama_models = []
# Start with predefined models from the shared source of truth
all_models = get_predefined_model_names()
# Fetch LiteLLM models (matches the /model command behavior)
try:
litellm_url = (
"https://raw.githubusercontent.com/BerriAI/litellm/main/"
"model_prices_and_context_window.json"
)
response = requests.get(litellm_url, timeout=2)
if response.status_code == 200:
litellm_data = response.json()
# Add LiteLLM models (sorted for consistency)
litellm_models = sorted(litellm_data.keys())
all_models.extend(litellm_models)
except Exception: # pylint: disable=broad-except
# Silently fail if LiteLLM is not available
pass
# Fetch Ollama models
try:
# Get Ollama models with a short timeout to prevent hanging
api_base = get_ollama_api_base()
@ -113,47 +134,17 @@ class FuzzyCommandCompleter(Completer):
# Fallback for older Ollama versions
models = data.get('items', [])
ollama_models = [(model.get('name', ''), []) for model in models]
ollama_models = [model.get('name', '') for model in models]
all_models.extend(ollama_models)
except Exception: # pylint: disable=broad-except
# Silently fail if Ollama is not available
pass
# Standard models always available
standard_models = [
# Alias models
"alias0",
# Claude 3.7 models
"claude-3-7-sonnet-20250219",
# Claude 3.5 models
"claude-3-5-sonnet-20240620",
"claude-3-5-20241122",
# Claude 3 models
"claude-3-opus-20240229",
"claude-3-sonnet-20240229",
"claude-3-haiku-20240307",
# OpenAI O-series models
"o1",
"o1-mini",
"o3-mini",
# OpenAI GPT models
"gpt-4o",
"gpt-4-turbo",
"gpt-3.5-turbo",
# DeepSeek models
"deepseek-v3",
"deepseek-r1"
]
# Combine standard models with Ollama models
self._cached_models = standard_models + ollama_models
# Cache all models that the /model command can handle
self._cached_models = all_models
# Create number mappings for models (1-based indexing)
# This matches the /model command numbering exactly
self._cached_model_numbers = {}
for i, model in enumerate(self._cached_models, 1):
self._cached_model_numbers[str(i)] = model
@ -484,7 +475,7 @@ class FuzzyCommandCompleter(Completer):
words = text.split()
# Refresh Ollama models periodically
self.fetch_ollama_models()
self.fetch_all_models()
if not text:
# Show all main commands with descriptions

257
src/cai/repl/commands/model.py
View File

@ -5,7 +5,7 @@ This module provides commands for viewing and changing the current LLM model.
import os
import datetime
# Standard library imports
from typing import List, Optional # Dict and Any removed as unused
from typing import List, Optional, Dict, Any
# Third-party imports
import requests # pylint: disable=import-error
@ -23,6 +23,138 @@ LITELLM_URL = (
)
def get_predefined_model_categories() -> Dict[str, List[Dict[str, str]]]:
"""Get the predefined model categories as the single source of truth.
This function serves as the authoritative source for all available models
across the CAI system. Other modules should import and use this function
to ensure consistency.
Returns:
Dictionary mapping category names to lists of model dictionaries
"""
return {
"Alias": [
{
"name": "alias0",
"description": (
"Best model for Cybersecurity AI tasks"
)
},
{
"name": "alias0-fast",
"description": (
"Fast version of alias0 for quick tasks"
)
}
],
"Anthropic Claude": [
{
"name": "claude-sonnet-4-20250514",
"description": (
"Excellent balance of performance and efficiency"
)
},
{
"name": "claude-3-7-sonnet-20250219",
"description": (
"Excellent model for complex reasoning and creative tasks"
)
},
{
"name": "claude-3-5-sonnet-20240620",
"description": (
"Excellent balance of performance and efficiency"
)
},
{
"name": "claude-3-5-haiku-20240307",
"description": (
"Fast and efficient model"
)
},
],
"OpenAI": [
{
"name": "o3-mini",
"description": "Latest mini model in the O-series"
},
{
"name": "gpt-4o",
"description": (
"Latest GPT-4 model with improved capabilities"
)
},
],
"DeepSeek": [
{
"name": "deepseek-v3",
"description": "DeepSeek's latest general-purpose model"
},
{
"name": "deepseek-r1",
"description": "DeepSeek's specialized reasoning model"
}
]
}
def get_all_predefined_models() -> List[Dict[str, Any]]:
"""Get all predefined models as a flat list with enriched data.
Returns:
List of model dictionaries with name, provider, category, description, and pricing
"""
model_categories = get_predefined_model_categories()
all_models = []
# Simple mapping from category to provider name
category_to_provider = {
"Alias": "OpenAI", # Alias models use OpenAI as base
"Anthropic Claude": "Anthropic",
"OpenAI": "OpenAI",
"DeepSeek": "DeepSeek"
}
for category, models in model_categories.items():
provider = category_to_provider.get(category, "Unknown")
for model in models:
# Get pricing info using COST_TRACKER
input_cost_per_token, output_cost_per_token = COST_TRACKER.get_model_pricing(model["name"])
# Convert to dollars per million tokens
input_cost_per_million = None
output_cost_per_million = None
if input_cost_per_token is not None and input_cost_per_token > 0:
input_cost_per_million = input_cost_per_token * 1000000
if output_cost_per_token is not None and output_cost_per_token > 0:
output_cost_per_million = output_cost_per_token * 1000000
all_models.append({
"name": model["name"],
"provider": provider,
"category": category,
"description": model["description"],
"input_cost": input_cost_per_million,
"output_cost": output_cost_per_million
})
return all_models
def get_predefined_model_names() -> List[str]:
"""Get a simple list of all predefined model names.
This is useful for autocompletion and simple model name lists.
Returns:
List of model name strings
"""
return [model["name"] for model in get_all_predefined_models()]
class ModelCommand(Command):
"""Command for viewing and changing the current LLM model."""
@ -64,108 +196,28 @@ class ModelCommand(Command):
Returns:
bool: True if the model was changed successfully
"""
# Define model categories and their models for easy reference
# Get all predefined models from the shared source of truth
# pylint: disable=invalid-name
MODEL_CATEGORIES = {
"Alias": [
{
"name": "alias0",
"description": (
"Best model for Cybersecurity AI tasks"
)
},
{
"name": "alias0-fast",
"description": (
"Fast version of alias0 for quick tasks"
)
}
],
"Anthropic Claude": [
{
"name": "claude-sonnet-4-20250514",
"description": (
"Excellent balance of performance and efficiency"
)
},
{
"name": "claude-3-7-sonnet-20250219",
"description": (
"Excellent model for complex reasoning and creative tasks"
)
},
{
"name": "claude-3-5-sonnet-20240620",
"description": (
"Excellent balance of performance and efficiency"
)
},
{
"name": "claude-3-5-haiku-20240307",
"description": (
"Fast and efficient model"
)
},
],
"OpenAI": [
{
"name": "o3-mini",
"description": "Latest mini model in the O-series"
},
{
"name": "gpt-4o",
"description": (
"Latest GPT-4 model with improved capabilities"
)
},
],
"DeepSeek": [
{
"name": "deepseek-v3",
"description": "DeepSeek's latest general-purpose model"
},
{
"name": "deepseek-r1",
"description": "DeepSeek's specialized reasoning model"
}
]
}
ALL_MODELS = get_all_predefined_models()
# Create a flat list of all models for numeric selection
# pylint: disable=invalid-name
ALL_MODELS = []
for category, models in MODEL_CATEGORIES.items():
for model in models:
# Get pricing info using COST_TRACKER
input_cost_per_token, output_cost_per_token = COST_TRACKER.get_model_pricing(model["name"])
# Also fetch LiteLLM model names to make numbering consistent with /model-show
litellm_model_names = []
try:
response = requests.get(LITELLM_URL, timeout=5)
if response.status_code == 200:
litellm_data = response.json()
# Add LiteLLM model names (sorted for consistency with /model-show)
litellm_model_names = sorted(litellm_data.keys())
except Exception: # pylint: disable=broad-except
# Silently fail if LiteLLM is not available
pass
# Convert to dollars per million tokens
input_cost_per_million = None
output_cost_per_million = None
if input_cost_per_token is not None and input_cost_per_token > 0:
input_cost_per_million = input_cost_per_token * 1000000
if output_cost_per_token is not None and output_cost_per_token > 0:
output_cost_per_million = output_cost_per_token * 1000000
ALL_MODELS.append({
"name": model["name"],
"provider": (
"Anthropic" if "claude" in model["name"]
else "DeepSeek" if "deepseek" in model["name"]
else "OpenAI"
),
"category": category,
"description": model["description"],
"input_cost": input_cost_per_million,
"output_cost": output_cost_per_million
})
# Update cached models
self.cached_models = [model["name"] for model in ALL_MODELS]
# Update cached models to include all models for number selection (consistent with /model-show)
predefined_model_names = [model["name"] for model in ALL_MODELS]
self.cached_models = predefined_model_names + litellm_model_names
self.cached_model_numbers = {
str(i): model["name"]
for i, model in enumerate(ALL_MODELS, 1)
str(i): model_name
for i, model_name in enumerate(self.cached_models, 1)
}
if not args: # pylint: disable=too-many-nested-blocks
@ -198,7 +250,7 @@ class ModelCommand(Command):
justify="right")
model_table.add_column("Description", style="white")
# Add all predefined models with numbers
# Add predefined models with numbers
for i, model in enumerate(ALL_MODELS, 1):
# Format pricing info as dollars per million tokens
input_cost_str = (
@ -242,7 +294,8 @@ class ModelCommand(Command):
ollama_models = data.get('items', [])
# Add Ollama models to the table with continuing numbers
start_index = len(ALL_MODELS) + 1
# (after predefined models + LiteLLM models in numbering)
start_index = len(predefined_model_names) + len(litellm_model_names) + 1
for i, model in enumerate(ollama_models, start_index):
model_name = model.get('name', '')
model_size = model.get('size', 0)
@ -275,7 +328,7 @@ class ModelCommand(Command):
self.cached_model_numbers[str(i)] = model_name
except Exception: # pylint: disable=broad-except
# Add a note about Ollama if we couldn't fetch models
start_index = len(ALL_MODELS) + 1
start_index = len(predefined_model_names) + len(litellm_model_names) + 1
model_table.add_row(
str(start_index),
"llama3",