From b58fcbd43c8bed1ddc6fb952106df9aacd02311f Mon Sep 17 00:00:00 2001
From: Jun Siang Cheah <git@jscheah.me>
Date: Sun, 9 Feb 2025 08:09:44 +0800
Subject: [PATCH] update samba models to developer tier

---
 README.md                    | 26 ++++++++++---------
 src/pull_available_models.py | 49 ++++++------------------------------
 2 files changed, 21 insertions(+), 54 deletions(-)

diff --git a/README.md b/README.md
index e0474d4..5409b4c 100644
--- a/README.md
+++ b/README.md
@@ -355,18 +355,20 @@ This lists various services that provide free access or credits towards API-base
 <tr><td>Qwen2-VL 7B Instruct</td></tr>
 <tr><td>Qwen2.5 72B Instruct</td></tr>
 <tr><td>Qwen2.5 Coder 32B Instruct</td></tr>
-<tr><td rowspan="12"><a href="https://cloud.sambanova.ai/" target="_blank">SambaNova Cloud</a></td><td rowspan="12">$5 for 3 months</td><td>Llama 3.1 8B</td><td>30 requests/minute</td></tr>
-<tr><td>Llama 3.1 70B</td><td>20 requests/minute</td></tr>
-<tr><td>Llama 3.1 405B</td><td>10 requests/minute</td></tr>
-<tr><td>Llama 3.2 1B</td><td>30 requests/minute</td></tr>
-<tr><td>Llama 3.2 3B</td><td>30 requests/minute</td></tr>
-<tr><td>Llama 3.2 11B</td><td>10 requests/minute</td></tr>
-<tr><td>Llama 3.2 90B</td><td>1 requests/minute</td></tr>
-<tr><td>Llama 3.3 70B</td><td>20 requests/minute</td></tr>
-<tr><td>Llama Guard 3 8B</td><td>30 requests/minute</td></tr>
-<tr><td>Qwen 2.5 72B</td><td>20 requests/minute</td></tr>
-<tr><td>Qwen 2.5 Coder 32B</td><td>20 requests/minute</td></tr>
-<tr><td>QwQ 32B Preview</td><td>10 requests/minute</td></tr>
+<tr><td rowspan="14"><a href="https://cloud.sambanova.ai/" target="_blank">SambaNova Cloud</a></td><td rowspan="14">$5 for 3 months</td><td>Llama 3.1 405B</td><td></td></tr>
+<tr><td>Llama 3.1 70B</td><td></td></tr>
+<tr><td>Llama 3.1 8B</td><td></td></tr>
+<tr><td>Llama 3.2 11B Vision</td><td></td></tr>
+<tr><td>Llama 3.2 1B</td><td></td></tr>
+<tr><td>Llama 3.2 3B</td><td></td></tr>
+<tr><td>Llama 3.2 90B Vision</td><td></td></tr>
+<tr><td>Llama 3.3 70B</td><td></td></tr>
+<tr><td>Llama-Guard-3-8B</td><td></td></tr>
+<tr><td>Qwen/QwQ-32B-Preview</td><td></td></tr>
+<tr><td>Qwen/Qwen2.5-72B-Instruct</td><td></td></tr>
+<tr><td>Qwen/Qwen2.5-Coder-32B-Instruct</td><td></td></tr>
+<tr><td>allenai/Llama-3.1-Tulu-3-405B</td><td></td></tr>
+<tr><td>deepseek-ai/DeepSeek-R1-Distill-Llama-70B</td><td></td></tr>
 <tr>
             <td><a href="https://studio.ai21.com/">AI21</a></td>
             <td>$10 for 3 months</td>
diff --git a/src/pull_available_models.py b/src/pull_available_models.py
index d2e5e31..6bb6a7a 100644
--- a/src/pull_available_models.py
+++ b/src/pull_available_models.py
@@ -657,54 +657,19 @@ def rate_limited_mistral_chat(client, **kwargs):
 
 def fetch_samba_models(logger):
     logger.info("Fetching SambaNova models...")
-    r = requests.get("https://community.sambanova.ai/t/rate-limits/321")
+    r = requests.get("https://cloud.sambanova.ai/api/pricing")
     r.raise_for_status()
-    prompt = f"""
-Here is the web page to extract data from:
-```html
-{r.text}
-```
-    """
-    logger.info("Extracting model rate limits from the provided web page...")
-    chat_response = rate_limited_mistral_chat(
-        mistral_client,
-        model="mistral-large-latest",
-        messages=[
-            {
-                "role": "system",
-                "content": """Extract the model rate limits as only integers from the provided web page into JSON:
-```json
-{
-  "Llama 3.1 405B": 10,
-  "Llama 3.2 90B": 1
-}
-```
-ONLY OUTPUT JSON!""",
-            },
-            {
-                "role": "user",
-                "content": prompt,
-            },
-        ],
-        response_format={
-            "type": "json_object",
-        },
-        temperature=0,
-    )
-    logger.info(chat_response.choices[0].message.content)
-    extracted_data = json.loads(chat_response.choices[0].message.content)
+    models = r.json()["prices"]
+    logger.info(f"Fetched {len(models)} models from SambaNova")
     ret_models = []
-    for model in extracted_data:
+    for model in models:
         ret_models.append(
             {
-                "id": model,
-                "name": model,
-                "limits": {
-                    "requests/minute": extracted_data[model],
-                },
+                "id": model["model_id"],
+                "name": model["model_name"],
             }
         )
-
+    ret_models = sorted(ret_models, key=lambda x: x["name"])
     return ret_models