Updated models

arena/c4.py

@@ -24,6 +24,8 @@ function refresh() {
 }
 """
 
+ALL_MODEL_NAMES = LLM.all_model_names()
+
 
 def message_html(game) -> str:
     """
@@ -181,11 +183,10 @@ def player_section(name, default):
     """
     Create the left and right sections of the UI
     """
-    all_model_names = LLM.all_model_names()
     with gr.Row():
         gr.HTML(f'<div style="text-align: center;font-size:18px">{name} Player</div>')
     with gr.Row():
-        dropdown = gr.Dropdown(all_model_names, value=default, label="LLM", interactive=True)
+        dropdown = gr.Dropdown(ALL_MODEL_NAMES, value=default, label="LLM", interactive=True)
     with gr.Row():
         gr.HTML('<div style="text-align: center;font-size:16px">Inner thoughts</div>')
     with gr.Row():
@@ -213,9 +214,7 @@ def make_display():
                     )
                 with gr.Row():
                     with gr.Column(scale=1):
-                        red_thoughts, red_dropdown = player_section(
-                            "Red", "openai/gpt-oss-120b via Groq"
-                        )
+                        red_thoughts, red_dropdown = player_section("Red", ALL_MODEL_NAMES[0])
                     with gr.Column(scale=2):
                         with gr.Row():
                             message = gr.HTML(
@@ -237,14 +236,14 @@ def make_display():
 
                     with gr.Column(scale=1):
                         yellow_thoughts, yellow_dropdown = player_section(
-                            "Yellow", "gemini-2.5-flash-lite"
+                            "Yellow", ALL_MODEL_NAMES[1]
                         )
             with gr.TabItem("Leaderboard") as leaderboard_tab:
                 with gr.Row():
                     with gr.Column(scale=1):
                         ratings_df = gr.Dataframe(
                             headers=["Player", "ELO"],
-                            label="Ratings",
+                            label="Ratings (recent models only)",
                             column_widths=[2, 1],
                             wrap=True,
                             col_count=2,

arena/game.py

@@ -3,6 +3,7 @@ from arena.player import Player
 from arena.record import get_games, Result, record_game, ratings
 from datetime import datetime
 from typing import List
+from arena.llm import LLM
 
 
 class Game:
@@ -54,9 +55,13 @@ class Game:
     @staticmethod
     def get_ratings():
         """
-        Return the ELO ratings of all players
+        Return the ELO ratings of all players - filter out any models that are not supported
         """
-        return ratings()
+        return {
+            model: rating
+            for model, rating in ratings().items()
+            if model in LLM.all_supported_model_names()
+        }
 
     def record(self):
         """

arena/llm.py

@@ -6,6 +6,9 @@ import logging
 from typing import Dict, Type, Self, List
 import os
 import time
+from dotenv import load_dotenv
+
+load_dotenv(override=True)
 
 logger = logging.getLogger(__name__)
 
@@ -110,17 +113,25 @@ class LLM(ABC):
                 mapping[model_name] = llm
         return mapping
 
+    @classmethod
+    def all_supported_model_names(cls) -> List[str]:
+        """
+        Return a list of all the model names supported by all subclasses of this one.
+        """
+        return list(cls.model_map().keys())
+
     @classmethod
     def all_model_names(cls) -> List[str]:
         """
         Return a list of all the model names supported.
         Use the ones specified in the model_map, but also check if there's an env variable set that restricts the models
         """
-        models = list(cls.model_map().keys())
+        models = cls.all_supported_model_names()
         allowed = os.getenv("MODELS")
+        print(f"Allowed models: {allowed}")
         if allowed:
             allowed_models = allowed.split(",")
-            return [model for model in models if model in allowed_models]
+            return [model for model in allowed_models if model in models]
         else:
             return models
 
@@ -144,10 +155,9 @@ class Claude(LLM):
     """
 
     model_names = [
-        "claude-3-5-sonnet-latest",
-        "claude-3-7-sonnet-latest",
         "claude-opus-4-1-20250805",
-        "claude-sonnet-4-5-20250929",
+        "claude-sonnet-4-5",
+        "claude-haiku-4-5",
     ]
 
     def __init__(self, model_name: str, temperature: float):
@@ -182,7 +192,7 @@ class GPT(LLM):
     A class to act as an interface to the remote AI, in this case GPT
     """
 
-    model_names = ["gpt-4o-mini", "gpt-4o", "gpt-5", "gpt-5-mini", "gpt-5-nano", "gpt-4.1-mini"]
+    model_names = ["gpt-5", "gpt-5-mini", "gpt-5-nano"]
 
     def __init__(self, model_name: str, temperature: float):
         """
@@ -199,7 +209,7 @@ class O1(LLM):
     A class to act as an interface to the remote AI, in this case O1
     """
 
-    model_names = ["o1-mini"]
+    model_names = []
 
     def __init__(self, model_name: str, temperature: float):
         """
@@ -231,7 +241,7 @@ class O3(LLM):
     A class to act as an interface to the remote AI, in this case O3
     """
 
-    model_names = ["o3-mini"]
+    model_names = []
 
     def __init__(self, model_name: str, temperature: float):
         """
@@ -269,8 +279,6 @@ class Gemini(LLM):
     """
 
     model_names = [
-        "gemini-2.0-flash",
-        "gemini-1.5-flash",
         "gemini-2.5-flash",
         "gemini-2.5-flash-lite",
         "gemini-2.5-pro",
@@ -347,7 +355,7 @@ class DeepSeekLocal(LLM):
     A class to act as an interface to the remote AI, in this case Ollama via the OpenAI client
     """
 
-    model_names = ["deepseek-r1:14b local"]
+    model_names = []
 
     def __init__(self, model_name: str, temperature: float):
         """
@@ -386,9 +394,6 @@ class GroqAPI(LLM):
     """
 
     model_names = [
-        "deepseek-r1-distill-llama-70b via Groq",
-        "llama-3.3-70b-versatile via Groq",
-        "mixtral-8x7b-32768 via Groq",
         "openai/gpt-oss-120b via Groq",
     ]