Update api_usage.py

api_usage.py

@@ -10,16 +10,29 @@ GPT_TYPES = ["gpt-3.5-turbo", "gpt-4", "gpt-4-32k"]
 #    "gpt-4": 500,
 #    "gpt-4-32k": 1000
 #}
+#RATE_LIMIT_PER_TIER_TURBO = {
+#    "free": 200,
+#    "tier-1-2-3": 3500,
+#    "tier-2-3": 5000,
+#    "tier-4-5": 10000
+#}
+#RATE_LIMIT_PER_TIER_GPT4 = {
+#    "tier-1": 500,
+#    "tier-2-3": 5000,
+#    "tier-4-5": 10000
+#}
+
 TOKEN_LIMIT_PER_TIER_TURBO = {
-    "free": 20000,
-    "tier-1": 40000,
+    "free": 40000,
+    "tier-1": 60000,
     "tier-1(old?)": 90000,
     "tier-2": 80000,
     "tier-3": 160000,
-    "tier-4-5": 1000000
+    "tier-4": 1000000,
+    "tier-5": 2000000
 }
 TOKEN_LIMIT_PER_TIER_GPT4 = {
-    "tier-free-1": 10000,
+    "tier-1": 10000,
     "tier-2": 40000,
     "tier-3": 80000,
     "tier-4-5": 300000
@@ -102,29 +115,30 @@ def format_status(list_models_avai, headers):
     tpm = []
     quota = ""
     for model in list_models_avai:
-        req_body = {"model": model, "messages": [{'role':'user', 'content': ''}], "max_tokens": 1}
+        req_body = {"model": model, "messages": [{'role':'user', 'content': ''}], "max_tokens": -0}
         r = requests.post(f"{BASE_URL}/chat/completions", headers=headers, json=req_body, timeout=10)
-        result = r.json()        
-        if "id" in result:
-            rpm_num = int(r.headers.get("x-ratelimit-limit-requests", 0))
-            tpm_num = int(r.headers.get('x-ratelimit-limit-tokens', 0))
-            _rpm = '{:,}'.format(rpm_num).replace(',', ' ')
-            _tpm = '{:,}'.format(tpm_num).replace(',', ' ')
-            _tpm_left = '{:,}'.format(int(r.headers.get('x-ratelimit-remaining-tokens', 0))).replace(',', ' ')
-            rpm.append(f"{_rpm} ({model})")
-            tpm.append(f"{_tpm} ({_tpm_left} left, {model})")
-            if model == GPT_TYPES[1]:
-                quota = check_key_tier(rpm_num, tpm_num, TOKEN_LIMIT_PER_TIER_GPT4, headers)
-            elif model == GPT_TYPES[0] and len(list_models_avai) == 1:
-                quota = check_key_tier(rpm_num, tpm_num, TOKEN_LIMIT_PER_TIER_TURBO, headers)
-            else:
-                continue
-        else:
+        result = r.json()
+        if "error" in result:
             e = result.get("error", {}).get("code", "")
-            rpm.append(f"0 ({model})")
-            tpm.append(f"0 ({model})")
-            quota = e
-            break
+            if e == None:
+                #print(r.headers)
+                rpm_num = int(r.headers.get("x-ratelimit-limit-requests", 0))
+                tpm_num = int(r.headers.get("x-ratelimit-limit-tokens_usage_based", 0))
+                tpm_left = int(r.headers.get("x-ratelimit-remaining-tokens_usage_based", 0))
+                rpm.append(f"{rpm_num} ({model})")
+                tpm.append(f"{tpm_num} ({tpm_left} left, {model})")
+                if model == GPT_TYPES[0]:
+                    quota = check_key_tier(tpm_num, TOKEN_LIMIT_PER_TIER_TURBO, headers)
+                #if model == GPT_TYPES[1]:
+                #    quota = check_key_tier(tpm_num, TOKEN_LIMIT_PER_TIER_GPT4, headers)
+                #elif model == GPT_TYPES[0] and len(list_models_avai) == 1:
+                #    quota = check_key_tier(tpm_num, TOKEN_LIMIT_PER_TIER_TURBO, headers)
+                #else:
+                #    continue
+            else:
+                rpm.append(f"0 ({model})")
+                tpm.append(f"0 ({model})")
+                quota = e                
     rpm_str = ""
     tpm_str = ""
     for i in range(len(rpm)):                
@@ -132,24 +146,19 @@ def format_status(list_models_avai, headers):
         tpm_str += tpm[i] + (", " if i < len(rpm)-1 else "")
     return rpm_str, tpm_str, quota
 
-def check_key_tier(rpm, tpm, dict, headers):
+def check_key_tier(rpm, dict, headers):
     dictItemsCount = len(dict)
     dictCount = 0
     for k, v in dict.items():        
-        if tpm == v:
-            if k == "tier-free-1":
-                if rpm == 500:
-                    return f"yes | tier-1"
-                else:
-                    return f"yes | free"
-            if k == "tier-4-5":
-                req_body = {"model": "text-embedding-ada-002", "input": "hiii"}
-                r = requests.post(f"{BASE_URL}/embeddings", headers=headers, json=req_body, timeout=10)
-                tpm_num = int(r.headers.get('x-ratelimit-limit-tokens', 0))
-                if tpm_num == 5000000:
-                    return f"yes | tier-4"
-                else:
-                    return f"yes | tier-5"
+        if rpm == v:
+            #if k == "tier-4-5":
+            #    req_body = {"model": "whisper-1"}
+            #    r = requests.post(f"{BASE_URL}/audio/transcriptions", headers=headers, json=req_body, timeout=10)
+            #    rpm_num = int(r.headers.get('x-ratelimit-limit-requests', 0))
+            #    if rpm_num == 100:
+            #        return f"yes | tier-4"
+            #    else:
+            #        return f"yes | tier-5"
             return f"yes | {k}"
         dictCount+=1
         if (dictCount == dictItemsCount):