4601 log model on response (#4781)

* add model tag to chatCompletion

* add modelTag `model` to async streaming
keeps default arguments for prompt token calculation where applied via explict arg

* fix HF default arg

* render all performance metrics as available for backward compatibility
add `timestamp` to both sync/async chat methods

* extract metrics string to function

frontend/src/components/WorkspaceChat/ChatContainer/ChatHistory/HistoricalMessage/Actions/RenderMetrics/index.jsx

@@ -1,3 +1,4 @@
+import { formatDateTimeAsMoment } from "@/utils/directories";
 import { numberWithCommas } from "@/utils/numbers";
 import React, { useEffect, useState, useContext } from "react";
 const MetricsContext = React.createContext();
@@ -41,6 +42,26 @@ function getAutoShowMetrics() {
   return window?.localStorage?.getItem(SHOW_METRICS_KEY) === "true";
 }
 
+/**
+ * Build the metrics string for a given metrics object
+ * - Model name
+ * - Duration and output TPS
+ * - Timestamp
+ * @param {metrics: {duration:number, outputTps: number, model?: string, timestamp?: number}} metrics
+ * @returns {string}
+ */
+function buildMetricsString(metrics = {}) {
+  return [
+    metrics?.model ? metrics.model : "",
+    `${formatDuration(metrics.duration)} (${formatTps(metrics.outputTps)} tok/s)`,
+    metrics?.timestamp
+      ? formatDateTimeAsMoment(metrics.timestamp, "MMM D, h:mm A")
+      : "",
+  ]
+    .filter(Boolean)
+    .join(" · ");
+}
+
 /**
  * Toggle the show metrics setting in localStorage `anythingllm_show_chat_metrics` key
  * @returns {void}
@@ -88,7 +109,7 @@ export function MetricsProvider({ children }) {
 
 /**
  * Render the metrics for a given chat, if available
- * @param {metrics: {duration:number, outputTps: number}} props
+ * @param {metrics: {duration:number, outputTps: number, model: string, timestamp: number}} props
  * @returns
  */
 export default function RenderMetrics({ metrics = {} }) {
@@ -110,8 +131,7 @@ export default function RenderMetrics({ metrics = {} }) {
       className={`border-none flex justify-end items-center gap-x-[8px] ${showMetricsAutomatically ? "opacity-100" : "opacity-0"} md:group-hover:opacity-100 transition-all duration-300`}
     >
       <p className="cursor-pointer text-xs font-mono text-theme-text-secondary opacity-50">
-        {formatDuration(metrics.duration)} ({formatTps(metrics.outputTps)}{" "}
+        {buildMetricsString(metrics)}
-        tok/s)
       </p>
     </button>
   );

server/utils/AiProviders/anthropic/index.js

@@ -171,6 +171,8 @@ class AnthropicLLM {
           total_tokens: promptTokens + completionTokens,
           outputTps: completionTokens / result.duration,
           duration: result.duration,
+          model: this.model,
+          timestamp: new Date(),
         },
       };
     } catch (error) {
@@ -190,7 +192,8 @@ class AnthropicLLM {
         temperature: Number(temperature ?? this.defaultTemp),
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/apipie/index.js

@@ -220,6 +220,8 @@ class ApiPieLLM {
         outputTps:
           (result.output.usage?.completion_tokens || 0) / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -237,7 +239,9 @@ class ApiPieLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/azureOpenAi/index.js

@@ -174,6 +174,8 @@ class AzureOpenAiLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -192,7 +194,9 @@ class AzureOpenAiLLM {
         n: 1,
         stream: true,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/bedrock/index.js

@@ -423,9 +423,7 @@ class AWSBedrockLLM {
             );
           }
           throw new Error(`AWSBedrock::getChatCompletion failed. ${e.message}`);
-        }),
+        })
-      messages,
-      false
     );
 
     const response = result.output;
@@ -450,6 +448,8 @@ class AWSBedrockLLM {
         total_tokens: response?.usage?.totalTokens ?? 0,
         outputTps: outputTps,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -492,7 +492,8 @@ class AWSBedrockLLM {
       const measuredStreamRequest = await LLMPerformanceMonitor.measureStream(
         stream,
         messages,
-        false // Indicate it's not a function call measurement
+        false,
+        this.model
       );
       return measuredStreamRequest;
     } catch (e) {

server/utils/AiProviders/cohere/index.js

@@ -124,6 +124,8 @@ class CohereLLM {
         total_tokens: promptTokens + completionTokens,
         outputTps: completionTokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -139,7 +141,8 @@ class CohereLLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/cometapi/index.js

@@ -225,6 +225,8 @@ class CometApiLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -242,7 +244,9 @@ class CometApiLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/deepseek/index.js

@@ -130,6 +130,8 @@ class DeepSeekLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -148,7 +150,8 @@ class DeepSeekLLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/dellProAiStudio/index.js

@@ -165,6 +165,8 @@ class DellProAiStudioLLM {
         total_tokens: result.output.usage?.total_tokens || 0,
         outputTps: result.output.usage?.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -182,7 +184,9 @@ class DellProAiStudioLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/fireworksAi/index.js

@@ -163,6 +163,8 @@ class FireworksAiLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -181,7 +183,8 @@ class FireworksAiLLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/foundry/index.js

@@ -224,6 +224,8 @@ class FoundryLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -242,7 +244,9 @@ class FoundryLLM {
         temperature,
         max_completion_tokens: this.promptWindowLimit(),
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/gemini/index.js

@@ -405,6 +405,8 @@ class GeminiLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -421,7 +423,8 @@ class GeminiLLM {
         },
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/genericOpenAi/index.js

@@ -193,6 +193,8 @@ class GenericOpenAiLLM {
         outputTps:
           (result.output?.usage?.completion_tokens || 0) / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -206,9 +208,10 @@ class GenericOpenAiLLM {
         temperature,
         max_tokens: this.maxTokens,
       }),
-      messages
+      messages,
       // runPromptTokenCalculation: true - There is not way to know if the generic provider connected is returning
-      // the correct usage metrics if any at all since any provider could be connected.
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/giteeai/index.js

@@ -170,6 +170,8 @@ class GiteeAILLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -183,7 +185,8 @@ class GiteeAILLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/groq/index.js

@@ -203,6 +203,8 @@ class GroqLLM {
           result.output.usage.completion_tokens /
           result.output.usage.completion_time,
         duration: result.output.usage.total_time,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -221,7 +223,8 @@ class GroqLLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/huggingface/index.js

@@ -117,6 +117,8 @@ class HuggingFaceLLM {
         outputTps:
           (result.output.usage?.completion_tokens || 0) / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -129,7 +131,9 @@ class HuggingFaceLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/koboldCPP/index.js

@@ -160,6 +160,8 @@ class KoboldCPPLLM {
         total_tokens: promptTokens + completionTokens,
         outputTps: completionTokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -173,7 +175,9 @@ class KoboldCPPLLM {
         temperature,
         max_tokens: this.maxTokens,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/liteLLM/index.js

@@ -154,6 +154,8 @@ class LiteLLM {
         outputTps:
           (result.output.usage?.completion_tokens || 0) / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -167,9 +169,10 @@ class LiteLLM {
         temperature,
         max_tokens: parseInt(this.maxTokens), // LiteLLM requires int
       }),
-      messages
+      messages,
       // runPromptTokenCalculation: true - We manually count the tokens because they may or may not be provided in the stream
-      // responses depending on LLM connected. If they are provided, then we counted for nothing, but better than nothing.
+      true,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/lmStudio/index.js

@@ -234,6 +234,8 @@ class LMStudioLLM {
         total_tokens: result.output.usage?.total_tokens || 0,
         outputTps: result.output.usage?.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -251,7 +253,9 @@ class LMStudioLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/localAi/index.js

@@ -145,6 +145,8 @@ class LocalAiLLM {
         total_tokens: promptTokens + completionTokens,
         outputTps: completionTokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -162,7 +164,9 @@ class LocalAiLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/mistral/index.js

@@ -139,6 +139,8 @@ class MistralLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -157,7 +159,8 @@ class MistralLLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/moonshotAi/index.js

@@ -136,6 +136,8 @@ class MoonshotAiLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -148,7 +150,9 @@ class MoonshotAiLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/novita/index.js

@@ -225,6 +225,8 @@ class NovitaLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -242,7 +244,9 @@ class NovitaLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/nvidiaNim/index.js

@@ -184,6 +184,8 @@ class NvidiaNimLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -201,7 +203,9 @@ class NvidiaNimLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/ollama/index.js

@@ -305,6 +305,8 @@ class OllamaAILLM {
         outputTps:
           result.output.usage.completion_tokens / result.output.usage.duration,
         duration: result.output.usage.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -326,7 +328,8 @@ class OllamaAILLM {
         },
       }),
       messages,
-      false
+      false,
+      this.model
     ).catch((e) => {
       throw this.#errorHandler(e);
     });

server/utils/AiProviders/openAi/index.js

@@ -175,6 +175,8 @@ class OpenAiLLM {
           ? usage.output_tokens / result.duration
           : 0,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -194,7 +196,8 @@ class OpenAiLLM {
         temperature: this.#temperature(this.model, temperature),
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/openRouter/index.js

@@ -276,6 +276,8 @@ class OpenRouterLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -300,13 +302,15 @@ class OpenRouterLLM {
         include_reasoning: true,
         user: user?.id ? `user_${user.id}` : "",
       }),
-      messages
+      messages,
       // We have to manually count the tokens
       // OpenRouter has a ton of providers and they all can return slightly differently
       // some return chunk.usage on STOP, some do it after stop, its inconsistent.
       // So it is possible reported metrics are inaccurate since we cannot reliably
       // catch the metrics before resolving the stream - so we just pretend this functionality
       // is not available.
+      true,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/perplexity/index.js

@@ -117,6 +117,8 @@ class PerplexityLLM {
         total_tokens: result.output.usage?.total_tokens || 0,
         outputTps: result.output.usage?.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -134,7 +136,9 @@ class PerplexityLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/ppio/index.js

@@ -176,6 +176,8 @@ class PPIOLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -193,7 +195,9 @@ class PPIOLLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/textGenWebUI/index.js

@@ -150,6 +150,8 @@ class TextGenWebUILLM {
         total_tokens: result.output.usage?.total_tokens || 0,
         outputTps: result.output.usage?.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -162,7 +164,9 @@ class TextGenWebUILLM {
         messages,
         temperature,
       }),
-      messages
+      messages,
+      true,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/togetherAi/index.js

@@ -209,6 +209,8 @@ class TogetherAiLLM {
         total_tokens: result.output.usage?.total_tokens || 0,
         outputTps: result.output.usage?.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -227,7 +229,8 @@ class TogetherAiLLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
     return measuredStreamRequest;
   }

server/utils/AiProviders/xai/index.js

@@ -147,6 +147,8 @@ class XAiLLM {
         total_tokens: result.output.usage.total_tokens || 0,
         outputTps: result.output.usage.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -165,7 +167,8 @@ class XAiLLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/AiProviders/zai/index.js

@@ -136,6 +136,8 @@ class ZAiLLM {
         total_tokens: result.output.usage?.total_tokens || 0,
         outputTps: result.output.usage?.completion_tokens / result.duration,
         duration: result.duration,
+        model: this.model,
+        timestamp: new Date(),
       },
     };
   }
@@ -149,7 +151,8 @@ class ZAiLLM {
         temperature,
       }),
       messages,
-      false
+      false,
+      this.model
     );
 
     return measuredStreamRequest;

server/utils/helpers/chat/LLMPerformanceMonitor.js

@@ -59,13 +59,15 @@ class LLMPerformanceMonitor {
    * Also attaches an `endMeasurement` method to the stream that will calculate the duration of the stream and metrics.
    * @param {Promise<OpenAICompatibleStream>} func
    * @param {Messages} messages - the messages sent to the LLM so we can calculate the prompt tokens since most providers do not return this on stream
-   * @param {boolean} runPromptTokenCalculation - whether to run the prompt token calculation to estimate the `prompt_tokens` metric. This is useful for providers that do not return this on stream.
+   * @param {boolean} runPromptTokenCalculation - [default: true] whether to run the prompt token calculation to estimate the `prompt_tokens` metric. This is useful for providers that do not return this on stream.
+   * @param {string} modelTag - the tag of the model that was used to generate the stream (eg: gpt-4o, claude-3-5-sonnet, qwen3/72b-instruct, etc.)
    * @returns {Promise<MonitoredStream>}
    */
   static async measureStream(
     func,
     messages = [],
-    runPromptTokenCalculation = true
+    runPromptTokenCalculation = true,
+    modelTag = ""
   ) {
     const stream = await func;
     stream.start = Date.now();
@@ -76,6 +78,7 @@ class LLMPerformanceMonitor {
       total_tokens: 0,
       outputTps: 0,
       duration: 0,
+      ...(modelTag ? { model: modelTag } : {}),
     };
 
     stream.endMeasurement = (reportedUsage = {}) => {
@@ -88,6 +91,7 @@ class LLMPerformanceMonitor {
         ...stream.metrics,
         ...reportedUsage,
         duration: reportedUsage?.duration ?? estimatedDuration,
+        timestamp: new Date(),
       };
 
       stream.metrics.total_tokens =