Skip to main content

Prism

Overview​

PropertyDetails
DescriptionPrism Inference serves open-weight models for coding agents over OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages APIs, with zero data retention on inputs and outputs.
Provider Route on LiteLLMprism/
Link to Provider DocPrism Documentation ↗
Base URLhttps://api.prisminference.com/v1
Supported Operations/chat/completions, /responses, /messages


We support ALL Prism models, just set prism/ as a prefix when sending requests

Available Models​

ModelDescriptionContext WindowMax Output
prism/deepseek-v4.1-flashDeepSeek-V4.1-Flash, the current generation for coding, vision, reasoning, and tool-use loops; takes text and image input1,000,000 tokens384,000 tokens
prism/deepseek-v4-flashDeepSeek-V4-Flash, a text-only model for fast coding and tool loops1,000,000 tokens384,000 tokens

Both models support reasoning, function calling, JSON mode, and JSON schema output, and LiteLLM ships their pricing (input, output, and cached input) so spend is tracked out of the box. Prism's catalog at GET https://api.prisminference.com/v1/models lists more models (for example glm-5.3 and kimi-k3); any of them works with the prism/ prefix, but for spend tracking on those you need to pass input_cost_per_token and output_cost_per_token in litellm_params until they are added to LiteLLM's model cost map.

Required Variables​

Environment Variables
os.environ["PRISM_API_KEY"] = ""  # your Prism API key

Usage - LiteLLM Python SDK​

Non-streaming​

Prism Non-streaming Completion
import os
import litellm
from litellm import completion

os.environ["PRISM_API_KEY"] = ""  # your Prism API key

messages = [{"content": "Hello, how are you?", "role": "user"}]

# Prism call
response = completion(
    model="prism/deepseek-v4.1-flash",
    messages=messages
)

print(response)

Streaming​

Prism Streaming Completion
import os
import litellm
from litellm import completion

os.environ["PRISM_API_KEY"] = ""  # your Prism API key

messages = [{"content": "Write a short story about AI", "role": "user"}]

# Prism call with streaming
response = completion(
    model="prism/deepseek-v4.1-flash",
    messages=messages,
    stream=True
)

for chunk in response:
    print(chunk)

Function Calling​

Prism Function Calling
import os
import litellm
from litellm import completion

os.environ["PRISM_API_KEY"] = ""  # your Prism API key

tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "Get the current weather in a location",
        "parameters": {
            "type": "object",
            "properties": {
                "city": {
                    "type": "string",
                    "description": "The city, e.g. San Francisco"
                }
            },
            "required": ["city"]
        }
    }
}]

messages = [{"role": "user", "content": "What's the weather in San Francisco?"}]

response = completion(
    model="prism/deepseek-v4.1-flash",
    messages=messages,
    tools=tools,
    tool_choice="auto"
)

print(response)

Structured Output​

Prism JSON Schema Output
import os
from litellm import completion

os.environ["PRISM_API_KEY"] = ""  # your Prism API key

response = completion(
    model="prism/deepseek-v4.1-flash",
    messages=[{"role": "user", "content": "The city is San Francisco"}],
    response_format={
        "type": "json_schema",
        "json_schema": {
            "name": "city",
            "strict": True,
            "schema": {
                "type": "object",
                "properties": {"city": {"type": "string"}},
                "required": ["city"],
                "additionalProperties": False,
            },
        },
    },
)

print(response)

Vision​

Image input is supported on prism/deepseek-v4.1-flash.

Prism Image Input
import os
from litellm import completion

os.environ["PRISM_API_KEY"] = ""  # your Prism API key

response = completion(
    model="prism/deepseek-v4.1-flash",
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "What colour fills this image?"},
            {"type": "image_url", "image_url": {"url": "https://example.com/image.png"}},
        ],
    }],
)

print(response)

Responses API​

Prism serves the OpenAI Responses API natively, so litellm.responses sends the request straight to https://api.prisminference.com/v1/responses.

Prism Responses API
import os
import litellm

os.environ["PRISM_API_KEY"] = ""  # your Prism API key

response = litellm.responses(
    model="prism/deepseek-v4.1-flash",
    input="Say hello",
)

print(response.output_text)

Anthropic Messages API​

Prism also serves the Anthropic Messages API natively, so litellm.anthropic.messages.acreate sends the request straight to https://api.prisminference.com/v1/messages.

Prism Anthropic Messages API
import asyncio
import os
import litellm

os.environ["PRISM_API_KEY"] = ""  # your Prism API key

async def main():
    response = await litellm.anthropic.messages.acreate(
        model="prism/deepseek-v4.1-flash",
        messages=[{"role": "user", "content": "Say hello"}],
        max_tokens=64,
    )
    print(response["content"][0]["text"])

asyncio.run(main())

Usage - LiteLLM Proxy Server​

config.yaml
model_list:
  - model_name: deepseek-v4.1-flash
    litellm_params:
      model: prism/deepseek-v4.1-flash
      api_key: os.environ/PRISM_API_KEY
  - model_name: deepseek-v4-flash
    litellm_params:
      model: prism/deepseek-v4-flash
      api_key: os.environ/PRISM_API_KEY

A deployment configured this way serves all three endpoints on the proxy: /v1/chat/completions, /v1/responses, and /v1/messages.

curl
curl http://0.0.0.0:4000/v1/chat/completions \
  -H "Authorization: Bearer $LITELLM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4.1-flash",
    "messages": [{"role": "user", "content": "Hello, how are you?"}]
  }'

Custom API Base​

Option 1: Environment variable

Custom API Base via env var
import os
from litellm import completion

os.environ["PRISM_API_BASE"] = "https://custom.prisminference.example/v1"
os.environ["PRISM_API_KEY"] = ""  # your API key

response = completion(
    model="prism/deepseek-v4.1-flash",
    messages=[{"content": "Hello!", "role": "user"}],
)

Option 2: Pass directly

Custom API Base via parameter
from litellm import completion

response = completion(
    model="prism/deepseek-v4.1-flash",
    messages=[{"content": "Hello!", "role": "user"}],
    api_base="https://custom.prisminference.example/v1",
    api_key="your-api-key",
)

Supported OpenAI Parameters​

  • temperature
  • max_tokens
  • max_completion_tokens
  • top_p
  • frequency_penalty
  • presence_penalty
  • stop
  • n
  • stream
  • stream_options
  • tools
  • tool_choice
  • response_format
  • seed
  • logprobs
  • top_logprobs
LiteLLM Enterprise
SSO/SAML, audit logs, spend tracking, multi-team management, and guardrails, built for production.
Learn more →