Added support for Llamafile and OpenAI

trustgraph/metering/pricelist.py

@@ -75,5 +75,30 @@ price_list = {
       "input_price": 0,
       "output_price": 0
     },
+    {
+      "model_name": "llama.cpp",
+      "input_price": 0,
+      "output_price": 0
+    },
+    {
+      "model_name": "gpt-4o",
+      "input_price": 0.000005,
+      "output_price": 0.000015
+    },
+    {
+      "model_name": "gpt-4o-2024-08-06",
+      "input_price": 0.0000025,
+      "output_price": 0.000010
+    },
+    {
+      "model_name": "gpt-4o-2024-05-13",
+      "input_price": 0.000005,
+      "output_price": 0.000015
+    },
+    {
+      "model_name": "gpt-4o-mini",
+      "input_price": 0.00000015,
+      "output_price": 0.0000006
+    },
   ]
 }

trustgraph/model/text_completion/llamafile/llm.py

@@ -107,12 +107,20 @@ class Processor(ConsumerProducer):
                     #}
                 )
 
+            inputtokens = resp.usage.prompt_tokens
+            outputtokens = resp.usage.completion_tokens
+
             print(resp.choices[0].message.content, flush=True)
+            print(f"Input Tokens: {inputtokens}", flush=True)
+            print(f"Output Tokens: {outputtokens}", flush=True)
 
             print("Send response...", flush=True)
             r = TextCompletionResponse(
                 response=resp.choices[0].message.content,
                 error=None,
+                in_token=inputtokens,
+                out_token=outputtokens,
+                model="llama.cpp"
             )
             self.send(r, properties={"id": id})
 
@@ -130,6 +138,9 @@ class Processor(ConsumerProducer):
                     message = str(e),
                 ),
                 response=None,
+                in_token=None,
+                out_token=None,
+                model=None,
             )
 
             self.producer.send(r, properties={"id": id})
@@ -148,6 +159,9 @@ class Processor(ConsumerProducer):
                     message = str(e),
                 ),
                 response=None,
+                in_token=None,
+                out_token=None,
+                model=None,
             )
 
             self.producer.send(r, properties={"id": id})

trustgraph/model/text_completion/openai/llm.py

@@ -108,13 +108,20 @@ class Processor(ConsumerProducer):
                         "type": "text"
                     }
                 )
-
+            
+            inputtokens = resp.usage.prompt_tokens
+            outputtokens = resp.usage.completion_tokens
             print(resp.choices[0].message.content, flush=True)
+            print(f"Input Tokens: {inputtokens}", flush=True)
+            print(f"Output Tokens: {outputtokens}", flush=True)
 
             print("Send response...", flush=True)
             r = TextCompletionResponse(
                 response=resp.choices[0].message.content,
                 error=None,
+                in_token=inputtokens,
+                out_token=outputtokens,
+                model=self.model
             )
             self.send(r, properties={"id": id})
 
@@ -132,6 +139,9 @@ class Processor(ConsumerProducer):
                     message = str(e),
                 ),
                 response=None,
+                in_token=None,
+                out_token=None,
+                model=None,
             )
 
             self.producer.send(r, properties={"id": id})
@@ -150,6 +160,9 @@ class Processor(ConsumerProducer):
                     message = str(e),
                 ),
                 response=None,
+                in_token=None,
+                out_token=None,
+                model=None,
             )
 
             self.producer.send(r, properties={"id": id})