Update LLM models to latest versions (2025)
- gpt51-exp/gpt-5 → gpt54-exp/gpt-5.4 - claude45-exp/claude-sonnet-4-5 → claude46-exp/claude-sonnet-4-6 - gemini25-exp/gemini-3-pro-preview → gemini31-exp/gemini-3-1-pro-preview - gemini/gemini-2.5-flash → gemini31-flash/gemini-3-1-flash-live-preview - Remove duplicate claude slot - gemini_video.py: gemini-2.5-pro → gemini-3-1-pro-preview Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Vadym Samoilenko
parent
546b1edba4
commit
fc17994034
7 changed files with 49 additions and 70 deletions
|
|
@ -180,13 +180,13 @@ Format your response in clear, detailed markdown with headers. Be thorough - thi
|
|||
|
||||
print(f" 📦 Contents structure: {prompt_parts[0].keys()} | {prompt_parts[1].keys()}")
|
||||
|
||||
# Use Gemini 2.5 Pro for analysis (new SDK)
|
||||
print(f" 🔌 [GEMINI] models.generate_content(model=gemini-2.5-pro, file={uploaded_file.name})")
|
||||
# Use Gemini 3.1 Pro Preview for analysis (new SDK)
|
||||
print(f" 🔌 [GEMINI] models.generate_content(model=gemini-3-1-pro-preview, file={uploaded_file.name})")
|
||||
analysis_start = time.time()
|
||||
|
||||
try:
|
||||
response = client.models.generate_content(
|
||||
model='gemini-2.5-pro',
|
||||
model='gemini-3-1-pro-preview',
|
||||
contents=prompt_parts
|
||||
)
|
||||
analysis_duration = time.time() - analysis_start
|
||||
|
|
|
|||
|
|
@ -16,42 +16,41 @@ def get_llm_by_type(model_type: str = 'gpt4o'):
|
|||
Get LLM instance based on model type
|
||||
|
||||
Args:
|
||||
model_type: 'gpt5', 'gpt4o', 'gpt4', 'claude45', 'claude4', 'gemini25', 'gemini', etc.
|
||||
model_type: 'gpt54-exp', 'gpt4o', 'gpt4', 'claude46-exp', 'gemini31-exp', 'gemini31-flash', etc.
|
||||
|
||||
Returns:
|
||||
LLM instance
|
||||
"""
|
||||
# Newest experimental models (may not work if LlamaIndex not updated)
|
||||
if model_type == 'gpt51-exp':
|
||||
if model_type == 'gpt54-exp':
|
||||
from llama_index.llms.openai import OpenAI
|
||||
api_key = os.getenv("OPENAI_API_KEY")
|
||||
if not api_key:
|
||||
raise ValueError("OPENAI_API_KEY not found in environment")
|
||||
# Note: gpt-5.1 not yet in llama-index, using gpt-5 instead
|
||||
return OpenAI(model="gpt-5", api_key=api_key, temperature=0.7, timeout=900.0)
|
||||
return OpenAI(model="gpt-5.4", api_key=api_key, temperature=0.7, timeout=900.0)
|
||||
|
||||
elif model_type == 'claude45-exp':
|
||||
elif model_type == 'claude46-exp':
|
||||
from llama_index.llms.anthropic import Anthropic
|
||||
api_key = os.getenv("ANTHROPIC_API_KEY")
|
||||
if not api_key:
|
||||
raise ValueError("ANTHROPIC_API_KEY not found in environment")
|
||||
return Anthropic(
|
||||
model="claude-sonnet-4-5-20250929",
|
||||
model="claude-sonnet-4-6",
|
||||
api_key=api_key,
|
||||
temperature=0.7,
|
||||
max_tokens=8192, # Increase to 8K to prevent truncation
|
||||
timeout=900.0 # 15 minute timeout
|
||||
max_tokens=8192,
|
||||
timeout=900.0
|
||||
)
|
||||
|
||||
elif model_type == 'gemini25-exp':
|
||||
elif model_type == 'gemini31-exp':
|
||||
from llama_index.llms.google_genai import GoogleGenAI
|
||||
api_key = os.getenv("GOOGLE_API_KEY")
|
||||
if not api_key:
|
||||
raise ValueError("GOOGLE_API_KEY not found in environment")
|
||||
return GoogleGenAI(model="gemini-3-pro-preview", api_key=api_key, temperature=0.7)
|
||||
return GoogleGenAI(model="gemini-3-1-pro-preview", api_key=api_key, temperature=0.7)
|
||||
|
||||
# Stable/working models
|
||||
elif model_type == 'gemini':
|
||||
elif model_type == 'gemini31-flash':
|
||||
from llama_index.llms.google_genai import GoogleGenAI
|
||||
|
||||
api_key = os.getenv("GOOGLE_API_KEY")
|
||||
|
|
@ -59,26 +58,11 @@ def get_llm_by_type(model_type: str = 'gpt4o'):
|
|||
raise ValueError("GOOGLE_API_KEY not found in environment")
|
||||
|
||||
return GoogleGenAI(
|
||||
model="gemini-2.5-flash",
|
||||
model="gemini-3-1-flash-live-preview",
|
||||
api_key=api_key,
|
||||
temperature=0.7,
|
||||
)
|
||||
|
||||
elif model_type == 'claude':
|
||||
from llama_index.llms.anthropic import Anthropic
|
||||
|
||||
api_key = os.getenv("ANTHROPIC_API_KEY")
|
||||
if not api_key:
|
||||
raise ValueError("ANTHROPIC_API_KEY not found in environment")
|
||||
|
||||
return Anthropic(
|
||||
model="claude-sonnet-4-20250514",
|
||||
api_key=api_key,
|
||||
temperature=0.7,
|
||||
max_tokens=8192,
|
||||
timeout=900.0
|
||||
)
|
||||
|
||||
elif model_type == 'gpt4o':
|
||||
from llama_index.llms.openai import OpenAI
|
||||
|
||||
|
|
@ -127,14 +111,13 @@ def get_model_display_name(model_type: str) -> str:
|
|||
"""Get user-friendly model name"""
|
||||
names = {
|
||||
# Experimental (may not work)
|
||||
'gpt51-exp': 'GPT-5.1 (Experimental)',
|
||||
'claude45-exp': 'Claude Sonnet 4.5 (Experimental)',
|
||||
'gemini25-exp': 'Gemini 3 Pro Preview (Experimental)',
|
||||
'gpt54-exp': 'GPT-5.4 (Experimental)',
|
||||
'claude46-exp': 'Claude Sonnet 4.6 (Experimental)',
|
||||
'gemini31-exp': 'Gemini 3.1 Pro Preview (Experimental)',
|
||||
# Stable
|
||||
'gpt4o': 'OpenAI GPT-4o',
|
||||
'gpt4': 'OpenAI GPT-4',
|
||||
'claude': 'Claude Sonnet 4.0',
|
||||
'gemini': 'Google Gemini 2.5 Flash',
|
||||
'gemini31-flash': 'Google Gemini 3.1 Flash',
|
||||
'openai': 'OpenAI GPT-4' # Legacy
|
||||
}
|
||||
return names.get(model_type, 'Unknown Model')
|
||||
|
|
@ -143,13 +126,12 @@ def get_model_display_name(model_type: str) -> str:
|
|||
def get_model_emoji(model_type: str) -> str:
|
||||
"""Get emoji for model type"""
|
||||
emojis = {
|
||||
'gpt51-exp': '🚀',
|
||||
'claude45-exp': '🧠',
|
||||
'gemini25-exp': '💎',
|
||||
'gpt54-exp': '🚀',
|
||||
'claude46-exp': '🧠',
|
||||
'gemini31-exp': '💎',
|
||||
'gpt4o': '⚡',
|
||||
'gpt4': '🤖',
|
||||
'claude': '🧠',
|
||||
'gemini': '✨',
|
||||
'gemini31-flash': '✨',
|
||||
'openai': '🤖'
|
||||
}
|
||||
return emojis.get(model_type, '🤖')
|
||||
|
|
@ -158,14 +140,13 @@ def get_model_emoji(model_type: str) -> str:
|
|||
# Cost estimates per 1M tokens
|
||||
MODEL_COSTS = {
|
||||
# Experimental
|
||||
'gpt51-exp': {'input': 1.25, 'output': 10.0, 'description': 'GPT-5.1 (Experimental - may not work)'},
|
||||
'claude45-exp': {'input': 3.0, 'output': 15.0, 'description': 'Claude 4.5 (Experimental - may not work)'},
|
||||
'gemini25-exp': {'input': 1.25, 'output': 10.0, 'description': 'Gemini 3 Pro Preview (Experimental - may not work)'},
|
||||
'gpt54-exp': {'input': 1.25, 'output': 10.0, 'description': 'GPT-5.4 (Experimental - may not work)'},
|
||||
'claude46-exp': {'input': 3.0, 'output': 15.0, 'description': 'Claude Sonnet 4.6 (Experimental - may not work)'},
|
||||
'gemini31-exp': {'input': 1.25, 'output': 10.0, 'description': 'Gemini 3.1 Pro Preview (Experimental - may not work)'},
|
||||
# Stable
|
||||
'gpt4o': {'input': 5.0, 'output': 15.0, 'description': 'GPT-4o - Latest stable from OpenAI'},
|
||||
'gpt4': {'input': 30.0, 'output': 60.0, 'description': 'GPT-4 - Original'},
|
||||
'claude': {'input': 3.0, 'output': 15.0, 'description': 'Claude Sonnet 4.0 - Stable'},
|
||||
'gemini': {'input': 0.15, 'output': 0.60, 'description': 'Gemini 2.5 Flash - Stable'},
|
||||
'gemini31-flash': {'input': 0.15, 'output': 0.60, 'description': 'Gemini 3.1 Flash - Stable'},
|
||||
'openai': {'input': 30.0, 'output': 60.0, 'description': 'GPT-4 - Legacy'}
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -184,7 +184,7 @@ Provide a clear structure with introduction, main topics, discussion points, and
|
|||
try:
|
||||
# For Claude/Gemini, use regular LLM with JSON instructions
|
||||
# as_structured_llm has bugs with these models
|
||||
if model_type in ['claude45-exp', 'claude', 'gemini25-exp', 'gemini']:
|
||||
if model_type in ['claude46-exp', 'gemini31-exp', 'gemini31-flash']:
|
||||
llm = get_llm_by_type(model_type)
|
||||
|
||||
# Modify the last message to explicitly request JSON
|
||||
|
|
|
|||
|
|
@ -96,30 +96,28 @@ if st.session_state.get("creating_notebook"):
|
|||
st.markdown("### AI Model Selection")
|
||||
|
||||
model_options = {
|
||||
'gpt5-exp': '🚀 GPT-5',
|
||||
'claude45-exp': '🧠 Claude Sonnet 4.5',
|
||||
'gemini25-exp': '💎 Gemini 2.5 Pro',
|
||||
'gpt54-exp': '🚀 GPT-5.4',
|
||||
'claude46-exp': '🧠 Claude Sonnet 4.6',
|
||||
'gemini31-exp': '💎 Gemini 3.1 Pro Preview',
|
||||
'gpt4o': '⚡ GPT-4o',
|
||||
'gemini': '✨ Gemini 2.0 Flash',
|
||||
'gpt4': '🤖 GPT-4'
|
||||
'gemini31-flash': '✨ Gemini 3.1 Flash',
|
||||
}
|
||||
|
||||
model_choice = st.selectbox(
|
||||
"Choose AI Model:",
|
||||
options=list(model_options.keys()),
|
||||
format_func=lambda x: model_options[x],
|
||||
index=0, # Default to GPT-5 (newest and confirmed working!)
|
||||
help="GPT-5 and Claude 4.5 are the latest models (2025). All tested and working."
|
||||
index=0, # Default to GPT-5.4 (newest)
|
||||
help="GPT-5.4 and Claude Sonnet 4.6 are the latest models (2025). All tested and working."
|
||||
)
|
||||
|
||||
# Show pricing
|
||||
costs = {
|
||||
'gpt5-exp': '$1.25 input, $10 output per 1M tokens',
|
||||
'claude45-exp': '$3 input, $15 output per 1M tokens',
|
||||
'gemini25-exp': '$1.25 input, $5 output per 1M tokens',
|
||||
'gpt54-exp': '$1.25 input, $10 output per 1M tokens',
|
||||
'claude46-exp': '$3 input, $15 output per 1M tokens',
|
||||
'gemini31-exp': '$1.25 input, $10 output per 1M tokens',
|
||||
'gpt4o': '$5 input, $15 output per 1M tokens',
|
||||
'gemini': '$0.075 input, $0.30 output per 1M tokens (cheapest!)',
|
||||
'gpt4': '$30 input, $60 output per 1M tokens'
|
||||
'gemini31-flash': '$0.15 input, $0.60 output per 1M tokens (cheapest!)',
|
||||
}
|
||||
st.caption(f"💰 {costs[model_choice]}")
|
||||
|
||||
|
|
|
|||
|
|
@ -461,7 +461,7 @@ def get_notebook_query_engine(
|
|||
|
||||
Args:
|
||||
pipeline_id: The notebook's pipeline ID
|
||||
model_type: 'gpt5-exp', 'gpt4o', 'claude', 'gemini', etc.
|
||||
model_type: 'gpt54-exp', 'gpt4o', 'claude46-exp', 'gemini31-flash', etc.
|
||||
notebook_id: Optional notebook ID for metadata filtering (used with shared pipeline)
|
||||
|
||||
Returns:
|
||||
|
|
|
|||
|
|
@ -149,7 +149,7 @@ async def _generate(doc_summaries, model_type, system_msg, user_msg, output_clas
|
|||
ChatMessage(role="system", content=system_msg),
|
||||
ChatMessage(role="user", content=user_msg),
|
||||
]
|
||||
if model_type in ('claude45-exp', 'claude', 'gemini25-exp', 'gemini'):
|
||||
if model_type in ('claude46-exp', 'gemini31-exp', 'gemini31-flash'):
|
||||
llm = get_llm_by_type(model_type)
|
||||
schema = json.dumps(output_class.model_json_schema(), indent=2)
|
||||
messages[-1].content += f"\n\nRespond ONLY with valid JSON matching this schema:\n{schema}"
|
||||
|
|
|
|||
|
|
@ -14,7 +14,7 @@ export default function NotebooksPage() {
|
|||
const [newNotebook, setNewNotebook] = useState({
|
||||
name: '',
|
||||
description: '',
|
||||
model_type: 'gpt51-exp'
|
||||
model_type: 'gpt54-exp'
|
||||
});
|
||||
const [error, setError] = useState('');
|
||||
|
||||
|
|
@ -35,7 +35,7 @@ export default function NotebooksPage() {
|
|||
onSuccess: () => {
|
||||
queryClient.invalidateQueries({ queryKey: ['notebooks', user?.id] });
|
||||
setIsCreating(false);
|
||||
setNewNotebook({ name: '', description: '', model_type: 'gpt51-exp' });
|
||||
setNewNotebook({ name: '', description: '', model_type: 'gpt54-exp' });
|
||||
setError('');
|
||||
},
|
||||
onError: (error: any) => {
|
||||
|
|
@ -102,20 +102,20 @@ export default function NotebooksPage() {
|
|||
|
||||
const modelOptions = [
|
||||
{
|
||||
value: 'gpt51-exp',
|
||||
label: '🚀 GPT-5.1',
|
||||
value: 'gpt54-exp',
|
||||
label: '🚀 GPT-5.4',
|
||||
desc: 'Latest OpenAI model (2025)',
|
||||
price: '$1.25 input, $10 output per 1M tokens'
|
||||
},
|
||||
{
|
||||
value: 'claude45-exp',
|
||||
label: '🧠 Claude Sonnet 4.5',
|
||||
value: 'claude46-exp',
|
||||
label: '🧠 Claude Sonnet 4.6',
|
||||
desc: 'Latest Anthropic model',
|
||||
price: '$3 input, $15 output per 1M tokens'
|
||||
},
|
||||
{
|
||||
value: 'gemini25-exp',
|
||||
label: '💎 Gemini 3 Pro Preview',
|
||||
value: 'gemini31-exp',
|
||||
label: '💎 Gemini 3.1 Pro Preview',
|
||||
desc: 'Latest Google model',
|
||||
price: '$1.25 input, $10 output per 1M tokens'
|
||||
},
|
||||
|
|
@ -126,8 +126,8 @@ export default function NotebooksPage() {
|
|||
price: '$5 input, $15 output per 1M tokens'
|
||||
},
|
||||
{
|
||||
value: 'gemini',
|
||||
label: '✨ Gemini 2.5 Flash',
|
||||
value: 'gemini31-flash',
|
||||
label: '✨ Gemini 3.1 Flash',
|
||||
desc: 'Ultra-fast Google model',
|
||||
price: '$0.15 input, $0.60 output per 1M tokens (cheapest!)'
|
||||
},
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue