Python SDK for AI agent monitoring, LLM cost tracking, benchmarking, and more. Integrates with most LLMs and agent frameworks including CrewAI, Agno, OpenAI Agents SDK, Langchain, Autogen, AG2, and CamelAI
Python agentagentopsagents-sdkaianthropicautogencost-estimationcrewaievalsevaluation-metricsgroqlangchainllmmistralollamaopenaiopenai-agents
---
title: IO Intelligence
description: "Track and analyze your IO Intelligence API calls with AgentOps"
---
AgentOps seamlessly integrates with IO Intelligence's OpenAI‑compatible API, allowing you to track and analyze every request without changing your workflow.
## Installation
<CodeGroup>
```bash pip
pip install openai
```
</CodeGroup>
## Basic Usage
Initialize AgentOps at the beginning of your application. Then create an **OpenAI** client that points at IO Intelligence's endpoint:
<CodeGroup>
```python Basic Usage
import agentops
from openai import OpenAI
# Initialise AgentOps (tracks every request automatically)
agentops.init("<YOUR_AGENTOPS_API_KEY>")
# Create IO Intelligence client (just add base_url)
client = OpenAI(
api_key="<YOUR_IO_INTELLIGENCE_API_KEY>",
base_url="https://api.intelligence.io.solutions/api/v1/"
)
response = client.chat.completions.create(
model="meta-llama/Llama-3.3-70B-Instruct",
messages=[{"role": "user", "content": "Say this is a test!"}]
)
print(response.choices[0].message.content)
```
</CodeGroup>
## Model Limits & Daily Quotas
Below are the free daily limits for each model (tokens are counted across input *and* output):
| LLM Model Name | Daily Chat quote | Daily API quote | Daily Embeddings quote | Context Length |
| --- | --- | --- | --- | --- |
| deepseek-ai/DeepSeek-R1 | 1,000,000 tk | 500,000 tk | N/A | 128,
| deepseek-ai/DeepSeek-R1-Distill-Llama-70B | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| meta-llama/Llama-3.3-70B-Instruct | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| deepseek-ai/DeepSeek-R1-Distill-Qwen-32B | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| Qwen/QwQ-32B-Preview | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| databricks/dbrx-instruct | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| deepseek-ai/DeepSeek-R1-Distill-Llama-8B | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| deepseek-ai/DeepSeek-R1-Distill-Qwen-14B | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| deepseek-ai/DeepSeek-R1-Distill-Qwen-7B | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| microsoft/phi-4 | 1,000,000 tk | 500,000 tk | N/A | 16,000 tk |
| mistralai/Mistral-Large-Instruct-2411 | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| neuralmagic/Llama-3.1-Nemotron-70B-Instruct-HF-FP8-dynamic | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| google/gemma-2-9b-it | 1,000,000 tk | 500,000 tk | N/A | 8,000 tk |
| nvidia/AceMath-7B-Instruct | 1,000,000 tk | 500,000 tk | N/A | 4,000 tk |
| CohereForAI/aya-expanse-32b | 1,000,000 tk | 500,000 tk | N/A | 8,000 tk |
| Qwen/Qwen2.5-Coder-32B-Instruct | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| THUDM/glm-4-9b-chat | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| CohereForAI/c4ai-command-r-plus-08-2024 | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| tiiuae/Falcon3-10B-Instruct | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| NovaSky-AI/Sky-T1-32B-Preview | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| bespokelabs/Bespoke-Stratos-32B | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| netease-youdao/Confucius-o1-14B | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| Qwen/Qwen2.5-1.5B-Instruct | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| mistralai/Ministral-8B-Instruct-2410 | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| openbmb/MiniCPM3-4B | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| jinaai/ReaderLM-v2 | 1,000,000 tk | 500,000 tk | N/A | 512,000 tk |
| ibm-granite/granite-3.1-8b-instruct | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| microsoft/Phi-3.5-mini-instruct | 1,000,000 tk | 500,000 tk | N/A | 128,000 tk |
| ozone-ai/0x-lite | 1,000,000 tk | 500,000 tk | N/A | 32,000 tk |
| mixedbread-ai/mxbai-embed-large-v1 | N/A | N/A | 500,000 tk | 512 tk |
| meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 | 1,000,000 tk | 500,000 tk | N/A | 430,000 tk |
## Quick Test (cURL)
<CodeGroup>
```bash cURL
curl https://api.intelligence.io.solutions/api/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <YOUR_IO_INTELLIGENCE_API_KEY>" \
-d '{
"model": "meta-llama/Llama-3.3-70B-Instruct",
"messages": [{"role": "user", "content": "Say this is a test!"}],
"temperature": 0.7
}'
```
</CodeGroup>
## Streaming Completions
AgentOps also captures streaming responses:
<CodeGroup>
```python Streaming Example
import agentops
from openai import OpenAI
agentops.init("<YOUR_AGENTOPS_API_KEY>")
client = OpenAI(
api_key="<YOUR_IO_INTELLIGENCE_API_KEY>",
base_url="https://api.intelligence.io.solutions/api/v1/"
)
stream = client.chat.completions.create(
model="meta-llama/Llama-3.3-70B-Instruct",
messages=[{"role": "user", "content": "Stream please"}],
stream=True
)
for chunk in stream:
print(chunk.choices[0].delta.content, end="", flush=True)
```
</CodeGroup>
## Advanced: Tool Calls
All OpenAI‑style tool calls work the same way—AgentOps will record parameters and results automatically.
<CodeGroup>
```python Tool Calls
import agentops, json
from openai import OpenAI
agentops.init("<YOUR_AGENTOPS_API_KEY>")
client = OpenAI(
api_key="<YOUR_IO_INTELLIGENCE_API_KEY>",
base_url="https://api.intelligence.io.solutions/api/v1/"
)
def get_weather(location):
return f"The weather in {location} is sunny."
messages = [
{"role": "system", "content": "You are a weather bot."},
{"role": "user", "content": "What's the weather like in Boston?"}
]
response = client.chat.completions.create(
model="meta-llama/Llama-3.3-70B-Instruct",
messages=messages,
tools=[{
"type":"function",
"function":{
"name":"get_weather",
"description":"Get weather for a city",
"parameters":{
"type":"object",
"properties":{
"location":{"type":"string"}
},
"required":["location"]
}
}
}]
)
tool_calls = response.choices[0].message.tool_calls
for call in tool_calls:
if call.function.name == "get_weather":
function_response = get_weather(json.loads(call.function.arguments)["location"])
messages.append({"role":"tool","tool_call_id":call.id,"content":function_response})
follow_up = client.chat.completions.create(
model="meta-llama/Llama-3.3-70B-Instruct",
messages=messages
)
print(follow_up.choices[0].message.content)
```
</CodeGroup>
<script type="module" src="/scripts/github_stars.js"></script>
<script type="module" src="/scripts/scroll-img-fadein-animation.js"></script>
<script type="module" src="/scripts/button_heartbeat_animation.js"></script>
<script type="css" src="/styles/styles.css"></script>