Model Directory & Pricing
Transparent wholesale pricing per million tokens with automated prompt caching discounts. All open-weight models execute inside hardware-attested Intel TDX enclaves.
| Model ID | Routing Tier | Context | Input / 1M | Cached Input | Output / 1M |
|---|---|---|---|---|---|
| phala/deepseek-v3.2 | Hardware TEE | 131k | $0.27 | $0.030 | $1.10 |
| phala/deepseek-r1 | Hardware TEE | 131k | $0.55 | $0.060 | $2.19 |
| phala/gpt-oss-120b | Hardware TEE | 131k | $0.10 | $0.020 | $0.49 |
| phala/qwen-2.5-coder-32b | Hardware TEE | 131k | $0.12 | $0.020 | $0.35 |
| phala/kimi-k2.6 | Hardware TEE | 262k | $1.09 | $0.150 | $4.60 |
| phala/glm-5.3 | Hardware TEE | 1000k | $1.40 | $0.140 | $4.40 |
| phala/llama-3.3-70b-instruct | Hardware TEE | 131k | $0.18 | $0.030 | $0.60 |
| black-forest-labs/flux-1-schnell | Hardware TEE | 0k | $5000.00 | $5000.000 | $0.00 |
| black-forest-labs/flux-1-dev | Hardware TEE | 0k | $25000.00 | $25000.000 | $0.00 |
| openai/dall-e-3 | Confidential Gateway | 0k | $40000.00 | $40000.000 | $0.00 |
| stabilityai/stable-diffusion-xl-base-1.0 | Confidential Gateway | 0k | $8000.00 | $8000.000 | $0.00 |
| openai/text-embedding-3-small | Confidential Gateway | 8k | $0.02 | $0.020 | $0.00 |
| openai/text-embedding-3-large | Confidential Gateway | 8k | $0.13 | $0.130 | $0.00 |
| voyageai/voyage-3 | Confidential Gateway | 32k | $0.12 | $0.120 | $0.00 |
| anthropic/claude-3-7-sonnet | Confidential Gateway | 200k | $3.00 | $0.300 | $15.00 |
| anthropic/claude-3-5-sonnet | Confidential Gateway | 200k | $3.00 | $0.300 | $15.00 |
| openai/gpt-4.5 | Confidential Gateway | 128k | $5.00 | $1.250 | $20.00 |
| openai/o3-mini | Confidential Gateway | 200k | $1.10 | $0.275 | $4.40 |
| openai/o1 | Confidential Gateway | 200k | $15.00 | $3.750 | $60.00 |
| openai/gpt-4o | Confidential Gateway | 128k | $2.50 | $0.625 | $10.00 |
| google/gemini-2.0-flash | Confidential Gateway | 1049k | $0.10 | $0.025 | $0.40 |
| google/gemini-2.0-pro | Confidential Gateway | 2097k | $1.25 | $0.313 | $5.00 |
OpenAI SDK Compatibility
Pass any model ID from the table above directly to client.chat.completions.create(model=...).