Hardware-Attested AI Compute.
Zero Accounts. Zero Logs.
Drop-in OpenAI-compatible inference with Intel TDX hardware enclaves, live attestation quotes you can verify yourself, and $10 flat 30-day TEE compute. No email, no KYC, no surveillance.
Why Sovereign Developers Switch to Phantom
How Phantom compares against OpenRouter, NanoGPT, Tinfoil, and Venice.
| Feature / Capability | Phantom (★) | OpenRouter | NanoGPT | Tinfoil.sh | Venice.ai |
|---|---|---|---|---|---|
| Hardware TEE Enclave Isolation | Intel TDX + NVIDIA CC | ❌ None (Plaintext Proxy) | ⚠️ Upstream (TDX) | ✅ AMD SEV-SNP + CC | ⚠️ Tier 1 Only |
| Live Hardware Attestation Quote | ✅ Raw TDX quote via API | ❌ None (56% Unknown) | ❌ Broken (Text only) | ✅ dm-verity Merkle | ❌ None |
| Per-Request Signed Response Receipt | ✅ Enclave receipt ID | ❌ None (Unsigned) | ⚠️ Text-Only ECDSA | ❌ Non-Transferable | ❌ None |
| Zero-Account Anonymous Checkout | ✅ SHA-256 Key Hash | ❌ Mandatory OAuth | ✅ Local Session | ❌ Mandatory Clerk Auth | ⚠️ Optional |
| Payment Rails & Deposit Surcharges | 0% Surcharge · XMR, USDC, Card | 5.5%+$0.80 Card · 5% USDC | 22 Cryptos, x402, Card | Stripe Cards Only (0% Crypto) | Card, Crypto, DIEM Stake |
| Flat-Rate Confidential TEE Plan | $10 / 30 Days (Daily Allowance) | ❌ No Subscriptions | ❌ TEE Excluded from $12 Sub | $20/mo (3M tok/hr cap) | $18–$200/mo (Credit-Capped) |
| Verifiable on Standard cURL / REST | ✅ 100% Native Headers | ❌ None | ❌ None | ❌ Fails (Proxy Required) | ❌ None |
Model Directory & Wholesale Rates
Wholesale per-token pricing with up to 90% prompt cache discounts.
| Model ID | Provider / Enclave | Context | Input / 1M | Cached Input | Output / 1M |
|---|---|---|---|---|---|
DeepSeek V3.2 deepseek/deepseek-v3.2 | 🛡️ Hardware TEE | 128k | $0.27 | $0.030 | $1.10 |
DeepSeek R1 deepseek/deepseek-v3.2 | 🛡️ Hardware TEE | 128k | $0.55 | $0.060 | $2.19 |
Claude 3.7 Sonnet anthropic/claude-sonnet-5 | ⚡ Frontier Gateway | 200k | $3.00 | $0.300 | $15.00 |
OpenAI o3-mini openai/o3-mini | ⚡ Frontier Gateway | 200k | $1.10 | $0.275 | $4.40 |
GPT-OSS 120B openai/gpt-oss-120b | 🛡️ Hardware TEE | 128k | $0.10 | $0.020 | $0.49 |
Gemini 2.0 Flash google/gemini-2.5-flash | ⚡ Frontier Gateway | 1M | $0.10 | $0.025 | $0.40 |
Qwen 2.5 Coder 32B qwen/qwen3-32b | 🛡️ Hardware TEE | 128k | $0.12 | $0.020 | $0.35 |
Kimi K2.6 moonshotai/kimi-k2.6 | 🛡️ Hardware TEE | 256k | $1.09 | $0.150 | $4.60 |
Perplexity Sonar perplexity/sonar | ⚡ Frontier Gateway | 128k | $1.00 | $0.250 | $1.00 |
Grok 2 (1212) x-ai/grok-2-1212 | ⚡ Frontier Gateway | 131k | $2.00 | $0.500 | $10.00 |
Codestral 2501 mistralai/codestral-2501 | ⚡ Frontier Gateway | 256k | $0.30 | $0.075 | $0.90 |
OpenAI GPT-4.5 openai/gpt-5.4 | ⚡ Frontier Gateway | 128k | $5.00 | $1.250 | $20.00 |
Drop-In OpenAI Compatibility
Switch models with 1 parameter change. Zero migrations, zero SDK lock-in.
from openai import OpenAI
# 1. Point standard OpenAI client to Phantom
client = OpenAI(
base_url="https://phantom.codes/v1",
api_key="sk-phantom-your-api-key"
)
# 2. Call any model (Claude 3.7, DeepSeek R1, GPT-4.5, o3-mini)
response = client.chat.completions.create(
model="deepseek/deepseek-v3.2",
messages=[{"role": "user", "content": "Explain KV prompt caching in 1 sentence."}],
stream=True
)
for chunk in response:
print(chunk.choices[0].delta.content or "", end="")Set OpenAI Base URL to https://phantom.codes/v1 with your sk-... key.
Why Developers Build on Phantom
High uptime, transparent token billing, and cryptographic privacy by default.
One Key for 15+ Models
Access DeepSeek, Claude 3.7, GPT-4.5, and o3-mini through a single API key without juggling multiple provider accounts or monthly subscriptions.
Hardware Enclave Privacy
Open-weight models execute in Intel TDX + NVIDIA CC enclaves. Live Intel TDX quotes are served straight from the enclave gateway so you can verify the platform and boot chain yourself.
Automatic Prompt Caching
Save up to 90% on long-context prompts, repository files, and system messages with automated KV prompt cache detection.