Together AI | Respan Docs

Set up Respan

Sign up — Create an account at platform.respan.ai
Create an API key — Generate one on the API keys page
Add credits or a provider key — Add credits on the Credits page or connect your own provider key on the Integrations page

Use AI

Add the Docs MCP to your AI coding tool to get help building with Respan. No API key needed.

1 {
2   "mcpServers": {
3     "respan-docs": {
4       "url": "https://docs.respan.ai/mcp"
5     }
6   }
7 }

What is Together AI?

The Together AI SDK is the official Python client for Together’s inference platform, providing access to open-source models like Llama, Mistral, and more. Respan can auto-instrument all Together AI calls for tracing, route them through the Respan gateway, or both.

Setup

Install packages

Tracing

Gateway

Both

$ pip install respan-ai opentelemetry-instrumentation-together together python-dotenv

Set environment variables

Tracing

Gateway

Both

$ export TOGETHER_API_KEY="YOUR_TOGETHER_API_KEY"
$ export RESPAN_API_KEY="YOUR_RESPAN_API_KEY"

Initialize and run

Tracing

Gateway

Both

1 import os
2 from dotenv import load_dotenv
3 
4 load_dotenv()
5 
6 from together import Together
7 from respan import Respan
8 
9 # Auto-discover and activate all installed instrumentors (Traceloop)
10 respan = Respan(is_auto_instrument=True)
11 
12 # Calls go directly to Together AI, auto-traced by Respan
13 client = Together()
14 
15 response = client.chat.completions.create(
16     model="meta-llama/Llama-3-70b-chat-hf",
17     messages=[{"role": "user", "content": "Say hello in three languages."}],
18 )
19 print(response.choices[0].message.content)
20 respan.flush()

View your trace

Open the Traces page to see your auto-instrumented LLM spans.

This step applies to Tracing and Both setups. The Gateway-only setup does not produce traces.

Configuration

Parameter	Type	Default	Description
`api_key`	`str \| None`	`None`	Falls back to `RESPAN_API_KEY` env var.
`base_url`	`str \| None`	`None`	Falls back to `RESPAN_BASE_URL` env var.
`is_auto_instrument`	`bool \| None`	`False`	Auto-discover and activate all installed instrumentors via OpenTelemetry entry points.
`customer_identifier`	`str \| None`	`None`	Default customer identifier for all spans.
`metadata`	`dict \| None`	`None`	Default metadata attached to all spans.
`environment`	`str \| None`	`None`	Environment tag (e.g. `"production"`).

Attributes

Attach customer identifiers, thread IDs, and metadata to spans.

In Respan()

Set defaults at initialization — these apply to all spans.

1 from respan import Respan
2 
3 respan = Respan(
4     is_auto_instrument=True,
5     customer_identifier="user_123",
6     metadata={"service": "chat-api", "version": "1.0.0"},
7 )

With propagate_attributes

Override per-request using a context manager.

1 from together import Together
2 from respan import Respan, workflow, propagate_attributes
3 
4 respan = Respan(
5     is_auto_instrument=True,
6     metadata={"service": "chat-api", "version": "1.0.0"},
7 )
8 client = Together()
9 
10 @workflow(name="handle_request")
11 def handle_request(user_id: str, question: str):
12     with propagate_attributes(
13         customer_identifier=user_id,
14         thread_identifier="conv_001",
15         metadata={"plan": "pro"},
16     ):
17         response = client.chat.completions.create(
18             model="meta-llama/Llama-3-70b-chat-hf",
19             messages=[{"role": "user", "content": question}],
20         )
21         print(response.choices[0].message.content)

Attribute	Type	Description
`customer_identifier`	`str`	Identifies the end user in Respan analytics.
`thread_identifier`	`str`	Groups related messages into a conversation.
`metadata`	`dict`	Custom key-value pairs. Merged with default metadata.

Decorators

Use @workflow and @task to create structured trace hierarchies.

1 from together import Together
2 from respan import Respan, workflow, task
3 
4 respan = Respan(is_auto_instrument=True)
5 client = Together()
6 
7 @task(name="generate_outline")
8 def outline(topic: str) -> str:
9     response = client.chat.completions.create(
10         model="meta-llama/Llama-3-70b-chat-hf",
11         messages=[
12             {"role": "user", "content": f"Create a brief outline about: {topic}"},
13         ],
14     )
15     return response.choices[0].message.content
16 
17 @workflow(name="content_pipeline")
18 def pipeline(topic: str):
19     plan = outline(topic)
20     response = client.chat.completions.create(
21         model="meta-llama/Llama-3-70b-chat-hf",
22         messages=[
23             {"role": "user", "content": f"Write content from this outline: {plan}"},
24         ],
25     )
26     print(response.choices[0].message.content)
27 
28 pipeline("Benefits of API gateways")
29 respan.flush()

Examples

Basic chat

1 response = client.chat.completions.create(
2     model="meta-llama/Llama-3-70b-chat-hf",
3     messages=[{"role": "user", "content": "Say hello in three languages."}],
4 )
5 print(response.choices[0].message.content)

Gateway features

The features below require the Gateway or Both setup from Step 3.

Switch models

Change the model parameter to use 250+ models from different providers through the same gateway.

1 # Together AI (Llama)
2 response = client.chat.completions.create(model="together_ai/meta-llama/Llama-3-70b-chat-hf", messages=messages)
3 
4 # OpenAI
5 response = client.chat.completions.create(model="gpt-4.1-nano", messages=messages)
6 
7 # Anthropic
8 response = client.chat.completions.create(model="claude-sonnet-4-5-20250929", messages=messages)

See the full model list.

Respan parameters

Pass additional Respan parameters via extra_body for gateway features.

1 response = client.chat.completions.create(
2     model="together_ai/meta-llama/Llama-3-70b-chat-hf",
3     messages=[{"role": "user", "content": "Hello"}],
4     extra_body={
5         "customer_identifier": "user_123",
6         "fallback_models": ["gpt-4.1-nano"],
7         "metadata": {"session_id": "abc123"},
8         "thread_identifier": "conversation_456",
9     },
10 )

See Respan parameters for the full list.