IntelCS-AI

Low-cost inference for open-weight models. We serve open models on efficient GPU capacity at some of the lowest per-token prices on the platform — with no compromise on features.

Why IntelCS-AI

• Price-first: open-weight models at floor prices — see the table below.
• Full feature parity: tool calling (function calling) and structured output (response_format: json_schema) on every conversational model.
• Long context: up to 1M tokens of context on supported models.
• Low latency: time-to-first-token well under the 5 s provider budget (measured ~0.9 s non-streaming).
• Autoscaling fleet: capacity scales out automatically with demand; routing, metering, and billing run on edge infrastructure.

Pricing

Our lineup rotates as we add capacity — check back for new models.

Usage

OpenAI-compatible, through the standard Hugging Face clients:

from huggingface_hub import InferenceClient

client = InferenceClient(
    model="google/gemma-3-4b-it",
    provider="intelcs-ai-iaas",
)

# Chat with tool calling
response = client.chat.completions.create(
    messages=[{"role": "user", "content": "What's the weather in Berlin?"}],
    tools=[{
        "type": "function",
        "function": {
            "name": "get_weather",
            "parameters": {"type": "object", "properties": {"location": {"type": "string"}}},
        },
    }],
)
print(response.choices[0].message.tool_calls)

# Structured output
response = client.chat.completions.create(
    messages=[{"role": "user", "content": "Extract the city: 'Flight to Berlin delayed'"}],
    response_format={"type": "json_schema", "json_schema": {"name": "city", "schema": {
        "type": "object", "properties": {"city": {"type": "string"}}, "required": ["city"],
    }}},
)
print(response.choices[0].message.content)  # {"city": "Berlin"}

Streaming (stream=True) is fully supported and metered per token.

Resources

• Website: intelcs.ai