Confidential AI Models
TEE models running inside Intel TDX enclaves at our inference provider. Requests go through our API.
11 TEE modelsOpenAI-compatibleIntel TDX
TEELLM
gemma-4-31B-turbo
890.1K runs / 7d
Input$0.24/M
Output$0.74/M
TEELLM
Qwen3.8-27B
Qwen
559.5K runs / 7d
Input$0.08/M
Output$2.50/M
TEELLM
Kimi-K2.6
Moonshot AI
92.9K runs / 7d
Input$0.78/M
Output$3.60/M
TEELLM
GLM-5.1
Z.ai
79.5K runs / 7d
Input$1.80/M
Output$5.80/M
TEELLM
DeepSeek-V3.2
DeepSeek
74.4K runs / 7d
Input$2.00/M
Output$2.00/M
TEELLM
Qwen3.6-27B
Qwen
57.8K runs / 7d
Input$0.50/M
Output$4.00/M
TEELLM
GLM-5.2
Z.ai
43.5K runs / 7d
Input$2.50/M
Output$7.90/M
TEELLM
Kimi-K3
Moonshot AI
19.0K runs / 7d
Input$3.70/M
Output$19.00/M
TEELLM
DeepSeek-V4-Flash-0731
DeepSeek
13.0K runs / 7d
Input$0.88/M
Output$2.64/M
TEELLM
Qwen3.5-27B
Qwen
10.4K runs / 7d
Input$0.40/M
Output$3.50/M
TEELLM
Qwen3.5-397B-A17B
Qwen
5.4K runs / 7d
Input$0.90/M
Output$6.00/M
API Quick Start
Use any OpenAI-compatible SDK. Your traffic travels over TLS to models running inside Intel TDX enclaves. Requests pass through our relay, which sees them in clear (TLS ends at our host). To keep plaintext off our servers, run your own inference server in a Confidential VM and terminate TLS inside its trust domain.
python
from openai import OpenAI
client = OpenAI(
base_url="https://api.voltagegpu.com/v1",
api_key="YOUR_API_KEY",
)
response = client.chat.completions.create(
model="google/gemma-4-31B-turbo-TEE",
messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)