feat: add reasoning effort to agent config model + support in core (#…

…9603)

Co-authored-by: Henry Fontanier <[email protected]>

core/src/blocks/chat.rs

@@ -290,6 +290,9 @@ impl Block for Chat {
                 if let Some(Value::String(s)) = v.get("response_format") {
                     extras["response_format"] = json!(s.clone());
                 }
+                if let Some(Value::String(s)) = v.get("reasoning_effort") {
+                    extras["reasoning_effort"] = json!(s.clone());
+                }
 
                 match extras.as_object().unwrap().keys().len() {
                     0 => None,

core/src/blocks/llm.rs

@@ -406,6 +406,9 @@ impl Block for LLM {
                 if let Some(Value::String(s)) = v.get("response_format") {
                     extras["response_format"] = json!(s.clone());
                 }
+                if let Some(Value::String(s)) = v.get("reasoning_effort") {
+                    extras["reasoning_effort"] = json!(s.clone());
+                }
 
                 match extras.as_object().unwrap().keys().len() {
                     0 => None,

core/src/providers/azure_openai.rs

@@ -447,8 +447,8 @@ impl LLM for AzureOpenAILLM {
             }
         }
 
-        let (openai_user, response_format) = match &extras {
-            None => (None, None),
+        let (openai_user, response_format, reasoning_effort) = match &extras {
+            None => (None, None, None),
             Some(v) => (
                 match v.get("openai_user") {
                     Some(Value::String(u)) => Some(u.to_string()),
@@ -458,6 +458,10 @@ impl LLM for AzureOpenAILLM {
                     Some(Value::String(f)) => Some(f.to_string()),
                     _ => None,
                 },
+                match v.get("reasoning_effort") {
+                    Some(Value::String(r)) => Some(r.to_string()),
+                    _ => None,
+                },
             ),
         };
 
@@ -501,6 +505,7 @@ impl LLM for AzureOpenAILLM {
                     },
                     response_format,
                     openai_user,
+                    reasoning_effort,
                     event_sender,
                 )
                 .await?
@@ -531,6 +536,7 @@ impl LLM for AzureOpenAILLM {
                         None => 0.0,
                     },
                     response_format,
+                    reasoning_effort,
                     openai_user,
                 )
                 .await?

core/src/providers/openai.rs

@@ -1014,6 +1014,7 @@ pub async fn streamed_chat_completion(
     presence_penalty: f32,
     frequency_penalty: f32,
     response_format: Option<String>,
+    reasoning_effort: Option<String>,
     user: Option<String>,
     event_sender: Option<UnboundedSender<Value>>,
 ) -> Result<(OpenAIChatCompletion, Option<String>)> {
@@ -1071,14 +1072,17 @@ pub async fn streamed_chat_completion(
     if tools.len() > 0 {
         body["tools"] = json!(tools);
     }
-    if tool_choice.is_some() {
+    if let Some(tool_choice) = tool_choice {
         body["tool_choice"] = json!(tool_choice);
     }
-    if response_format.is_some() {
+    if let Some(response_format) = response_format {
         body["response_format"] = json!({
-            "type": response_format.unwrap(),
+            "type": response_format,
         });
     }
+    if let Some(reasoning_effort) = reasoning_effort {
+        body["reasoning_effort"] = json!(reasoning_effort);
+    }
 
     let client = builder
         .body(body.to_string())
@@ -1436,6 +1440,7 @@ pub async fn chat_completion(
     presence_penalty: f32,
     frequency_penalty: f32,
     response_format: Option<String>,
+    reasoning_effort: Option<String>,
     user: Option<String>,
 ) -> Result<(OpenAIChatCompletion, Option<String>)> {
     let mut body = json!({
@@ -1449,7 +1454,7 @@ pub async fn chat_completion(
     if user.is_some() {
         body["user"] = json!(user);
     }
-    if model_id.is_some() {
+    if let Some(model_id) = model_id {
         body["model"] = json!(model_id);
     }
     if let Some(mt) = max_tokens {
@@ -1459,17 +1464,20 @@ pub async fn chat_completion(
         body["stop"] = json!(stop);
     }
 
-    if response_format.is_some() {
+    if let Some(response_format) = response_format {
         body["response_format"] = json!({
-            "type": response_format.unwrap(),
+            "type": response_format,
         });
     }
     if tools.len() > 0 {
         body["tools"] = json!(tools);
     }
-    if tool_choice.is_some() {
+    if let Some(tool_choice) = tool_choice {
         body["tool_choice"] = json!(tool_choice);
     }
+    if let Some(reasoning_effort) = reasoning_effort {
+        body["reasoning_effort"] = json!(reasoning_effort);
+    }
 
     let mut req = reqwest::Client::new()
         .post(uri.to_string())
@@ -2005,8 +2013,8 @@ impl LLM for OpenAILLM {
             }
         }
 
-        let (openai_org_id, openai_user, response_format) = match &extras {
-            None => (None, None, None),
+        let (openai_org_id, openai_user, response_format, reasoning_effort) = match &extras {
+            None => (None, None, None, None),
             Some(v) => (
                 match v.get("openai_organization_id") {
                     Some(Value::String(o)) => Some(o.to_string()),
@@ -2020,6 +2028,10 @@ impl LLM for OpenAILLM {
                     Some(Value::String(f)) => Some(f.to_string()),
                     _ => None,
                 },
+                match v.get("reasoning_effort") {
+                    Some(Value::String(r)) => Some(r.to_string()),
+                    _ => None,
+                },
             ),
         };
 
@@ -2065,6 +2077,7 @@ impl LLM for OpenAILLM {
                     None => 0.0,
                 },
                 response_format,
+                reasoning_effort,
                 openai_user,
                 event_sender.clone(),
             )
@@ -2095,6 +2108,7 @@ impl LLM for OpenAILLM {
                     None => 0.0,
                 },
                 response_format,
+                reasoning_effort,
                 openai_user,
             )
             .await?

core/src/providers/togetherai.rs

@@ -195,6 +195,7 @@ impl LLM for TogetherAILLM {
                 },
                 None,
                 None,
+                None,
                 event_sender.clone(),
             )
             .await?
@@ -225,6 +226,7 @@ impl LLM for TogetherAILLM {
                 },
                 None,
                 None,
+                None,
             )
             .await?
         };

front/lib/models/assistant/agent.ts

@@ -1,5 +1,6 @@
 import type {
   AgentConfigurationScope,
+  AgentReasoningEffort,
   AgentStatus,
   GlobalAgentStatus,
   ModelIdType,
@@ -34,6 +35,7 @@ export class AgentConfiguration extends BaseModel<AgentConfiguration> {
   declare providerId: ModelProviderIdType;
   declare modelId: ModelIdType;
   declare temperature: number;
+  declare reasoningEffort: AgentReasoningEffort | null;
 
   declare pictureUrl: string;
 
@@ -107,6 +109,10 @@ AgentConfiguration.init(
       allowNull: false,
       defaultValue: 0.7,
     },
+    reasoningEffort: {
+      type: DataTypes.STRING,
+      allowNull: true,
+    },
     maxStepsPerRun: {
       type: DataTypes.INTEGER,
       allowNull: true,

front/migrations/db/migration_136.sql

@@ -0,0 +1,2 @@
+-- Migration created on Dec 23, 2024
+ALTER TABLE "public"."agent_configurations" ADD COLUMN "reasoningEffort" VARCHAR(255);

types/src/front/assistant/agent.ts

@@ -176,10 +176,13 @@ export type AgentUsageType = {
 
 export type AgentRecentAuthors = readonly string[];
 
+export type AgentReasoningEffort = "low" | "medium" | "high";
+
 export type AgentModelConfigurationType = {
   providerId: ModelProviderIdType;
   modelId: ModelIdType;
   temperature: number;
+  reasoningEffort?: AgentReasoningEffort;
 };
 
 export type LightAgentConfigurationType = {
@@ -228,6 +231,8 @@ export type LightAgentConfigurationType = {
   //
   // Example: [[1,2], [3,4]] means (1 OR 2) AND (3 OR 4)
   requestedGroupIds: string[][];
+
+  reasoningEffort?: AgentReasoningEffort;
 };
 
 export type AgentConfigurationType = LightAgentConfigurationType & {