Merge v0.77.1

Author: FlyingFathead

README.md

@@ -240,6 +240,26 @@ If you run into any issues, consult the logs or reach out on the repository's [I
 ---
 
 # Changelog
+
+- v0.77.1 - OpenAI newer model (5.x series) / tools compatibility update
+  - Added support for modern OpenAI Chat Completions tool calling via `tools` / `tool_choice` / `tool_calls`.
+  - Added backward-compatible parsing for both modern `tool_calls` and legacy `function_call` responses.
+  - Added `build_openai_tools()` to convert the existing legacy ChatKeke `custom_functions` format into OpenAI's modern tools format.
+  - Added `extract_function_call_or_none()` so function-call detection no longer depends only on the old `message["function_call"]` field.
+  - Updated payload building so function calling can work with newer model families, including GPT-5.x-style models.
+  - Restricted `temperature` to GPT-4-family models only, avoiding API errors from newer reasoning-style models that reject sampling parameters.
+  - Uses `max_completion_tokens` for newer non-GPT-4 models while keeping `max_tokens` for GPT-4-family models.
+  - Hardened OpenAI response parsing to avoid confusing `KeyError: 'choices'` failures when the API returns an error body.
+  - Added safer handling for missing or empty `choices` responses.
+  - Updated function-call follow-up calls so tool results are appended into chat context and the final user-facing answer is generated with `include_functions=False`, reducing accidental recursive tool calls.
+  - Updated affected function branches:
+    - `calculate_expression`
+    - `get_weather`
+    - `get_duckduckgo_search`
+    - `get_website_dump`
+    - `get_stock_price`
+    - `query_perplexity`
+    - `manage_reminder`
 - v0.7616 - Changed default Perplexity API model to `sonar` as per new model changes in the Perplexity API
 - v0.7615 - Parsing improvements
   - Improved text formatting & escaping in complex markdown vs. html cases

config/config.ini

@@ -15,7 +15,7 @@ AskForTokenIfNotFound = True
 
 # Model to use via OpenAI API
 # NOTE: SEE ALSO THE NEW AUTO-SWITCHING FEATURE UNDER: [ModelAutoSwitch]
-Model = gpt-4o-mini
+Model = gpt-5.4-mini
 
 # Model temperature; OpenAI's default is 0.7
 # Higher temperature equals more creative (= less factual) output. much over 1.0 not recommended.
@@ -126,11 +126,11 @@ Enabled = True
 # The preferred, more capable model to use by default (e.g., gpt-4o, gpt-4.5-preview).
 # This model will be used until its daily token limit (PremiumTokenLimit) is reached.
 # PremiumModel = gpt-4o
-PremiumModel = gpt-4.1
+PremiumModel = gpt-5.4-mini
 
 # The cheaper model to switch to when the PremiumTokenLimit is reached (e.g., gpt-4o-mini).
 # This model has its own daily token limit (MiniTokenLimit).
-FallbackModel = gpt-4o-mini
+FallbackModel = gpt-5.4-mini
 
 # Daily token limit for models considered "Premium" (e.g., gpt-4o).
 # Set to number of tokens (i.e. 1000000 for 1M; 500000 for 500k etc)

src/main.py

@@ -6,7 +6,7 @@
 # https://github.com/FlyingFathead/TelegramBot-OpenAI-API
 # ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 # version of this program
-version_number = "0.7616"
+version_number = "0.77.1"
 
 # Add the project root directory to Python's path
 import sys