New OpenAI models: o4_mini, o3

Author: peterbanda

openai-client/src/main/scala/io/cequence/openaiscala/service/impl/OpenAIChatCompletionServiceImpl.scala

@@ -61,11 +61,18 @@ trait ChatCompletionBodyMaker {
     ModelId.o1_mini_2024_09_12
   )
 
-  private val regularO1Or3Models = Set(
+  private val regularOModels = Set(
     ModelId.o1,
     ModelId.o1_2024_12_17,
+    ModelId.o1_pro,
+    ModelId.o1_pro_2025_03_19,
+    ModelId.o3,
+    ModelId.o3_2025_04_16,
     ModelId.o3_mini,
-    ModelId.o3_mini_2025_01_31
+    ModelId.o3_mini_2025_01_31,
+    ModelId.o3_mini_high,
+    ModelId.o4_mini,
+    ModelId.o4_mini_2025_04_16
   )
 
   protected def createBodyParamsForChatCompletion(
@@ -84,12 +91,12 @@ trait ChatCompletionBodyMaker {
 
     val messageJsons = messagesFinal.map(Json.toJson(_)(messageWrites))
 
-    // O1/3 models need some special treatment... revisit this later
+    // regular O models need some special treatment... revisit this later
     val settingsFinal =
       if (o1PreviewModels.contains(settings.model))
         ChatCompletionSettingsConversions.o1Preview(settings)
-      else if (regularO1Or3Models.contains(settings.model))
-        ChatCompletionSettingsConversions.o1And3(settings)
+      else if (regularOModels.contains(settings.model))
+        ChatCompletionSettingsConversions.o(settings)
       else
         settings
 

openai-core/src/main/scala/io/cequence/openaiscala/domain/ModelId.scala

@@ -177,7 +177,12 @@ object ModelId {
   // 16k context, gpt_3_5_turbo will point to this model from Dec 11, 2023
   val gpt_3_5_turbo_1106 = "gpt-3.5-turbo-1106"
 
-  // Q*/Strawberry
+  // O models - Q*/Strawberry
+
+  val o4_mini = "o4-mini"
+  val o4_mini_2025_04_16 = "o4-mini-2025-04-16"
+  val o3 = "o3"
+  val o3_2025_04_16 = "o3-2025-04-16"
   val o3_mini = "o3-mini"
   val o3_mini_2025_01_31 = "o3-mini-2025-01-31"
   // High-compute version of o3-mini

openai-core/src/main/scala/io/cequence/openaiscala/service/OpenAIChatCompletionExtra.scala

@@ -192,13 +192,16 @@ object OpenAIChatCompletionExtra {
     ModelId.gpt_4o,
     ModelId.gpt_4o_2024_08_06,
     ModelId.gpt_4o_2024_11_20,
+    ModelId.o4_mini_2025_04_16,
+    ModelId.o3,
+    ModelId.o3_2025_04_16,
+    ModelId.o3_mini,
+    ModelId.o3_mini_high,
+    ModelId.o3_mini_2025_01_31,
     ModelId.o1,
     ModelId.o1_2024_12_17,
     ModelId.o1_pro,
     ModelId.o1_pro_2025_03_19,
-    ModelId.o3_mini,
-    ModelId.o3_mini_high,
-    ModelId.o3_mini_2025_01_31,
     NonOpenAIModelId.gemini_2_5_pro_preview_03_25,
     NonOpenAIModelId.gemini_2_5_pro_exp_03_25,
     NonOpenAIModelId.gemini_2_0_flash,

openai-core/src/main/scala/io/cequence/openaiscala/service/adapter/ChatCompletionSettingsConversions.scala

@@ -44,14 +44,14 @@ object ChatCompletionSettingsConversions {
           extra_params =
             settings.extra_params + ("max_completion_tokens" -> settings.max_tokens.get)
         ),
-      Some("O1/3 models don't support max_tokens, converting to max_completion_tokens")
+      Some("O models don't support max_tokens, converting to max_completion_tokens")
     ),
     // temperature
     FieldConversionDef(
       settings => settings.temperature.isDefined && settings.temperature.get != 1,
       _.copy(temperature = Some(1d)),
       Some(
-        "O1/3 models don't support temperature values other than the default of 1, converting to 1."
+        "O models don't support temperature values other than the default of 1, converting to 1."
       ),
       warning = true
     ),
@@ -60,7 +60,7 @@ object ChatCompletionSettingsConversions {
       settings => settings.top_p.isDefined && settings.top_p.get != 1,
       _.copy(top_p = Some(1d)),
       Some(
-        "O1/3 models don't support top p values other than the default of 1, converting to 1."
+        "O models don't support top p values other than the default of 1, converting to 1."
       ),
       warning = true
     ),
@@ -69,7 +69,7 @@ object ChatCompletionSettingsConversions {
       settings => settings.presence_penalty.isDefined && settings.presence_penalty.get != 0,
       _.copy(presence_penalty = Some(0d)),
       Some(
-        "O1/3 models don't support presence penalty values other than the default of 0, converting to 0."
+        "O models don't support presence penalty values other than the default of 0, converting to 0."
       ),
       warning = true
     ),
@@ -78,7 +78,7 @@ object ChatCompletionSettingsConversions {
       settings => settings.frequency_penalty.isDefined && settings.frequency_penalty.get != 0,
       _.copy(frequency_penalty = Some(0d)),
       Some(
-        "O1/3 models don't support frequency penalty values other than the default of 0, converting to 0."
+        "O models don't support frequency penalty values other than the default of 0, converting to 0."
       ),
       warning = true
     ),
@@ -87,7 +87,7 @@ object ChatCompletionSettingsConversions {
       settings => settings.parallel_tool_calls.isDefined,
       _.copy(parallel_tool_calls = None),
       Some(
-        "O1/3 models don't support parallel tool calls, converting to None."
+        "O models don't support parallel tool calls, converting to None."
       ),
       warning = true
     )
@@ -106,7 +106,7 @@ object ChatCompletionSettingsConversions {
         warning = true
       )
 
-  val o1And3: SettingsConversion = generic(oBaseConversions)
+  val o: SettingsConversion = generic(oBaseConversions)
 
   val o1Preview: SettingsConversion = generic(o1PreviewConversions)
 

openai-examples/src/main/scala/io/cequence/openaiscala/examples/CreateChatCompletion.scala

@@ -17,9 +17,9 @@ object CreateChatCompletion extends Example {
       .createChatCompletion(
         messages = messages,
         settings = CreateChatCompletionSettings(
-          model = ModelId.gpt_4_1,
-          temperature = Some(0),
-          max_tokens = Some(4000),
+          model = ModelId.o3,
+          max_tokens = Some(1000),
+          temperature = Some(0.1),
           service_tier = Some(ServiceTier.auto),
           metadata = Map()
         )