PHANTOM API

// 01

Quick start

▸ curl

$ curl https://phantom.codes/v1/chat/completions \
    -H "Authorization: Bearer YOUR_PHANTOM_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "phantom/qwen3-32b",
      "messages": [{"role":"user","content":"Hello"}]
    }'

▸ Python

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_PHANTOM_KEY",
    base_url="https://phantom.codes/v1",
)

r = client.chat.completions.create(
    model="phantom/deepseek-v4-flash",
    messages=[{"role": "user", "content": "Hello"}],
)
print(r.choices[0].message.content)

▸ Node

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "YOUR_PHANTOM_KEY",
  baseURL: "https://phantom.codes/v1",
});

const r = await client.chat.completions.create({
  model: "phantom/qwen3.5-122b-a10b",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(r.choices[0].message.content);

Need crypto? See /#pricing for accepted coins and how to acquire them.

// 02

Endpoints

▸ Check balance

$ curl https://phantom.codes/v1/key/balance \
    -H "Authorization: Bearer YOUR_PHANTOM_KEY"

{"balance_usd": 9.7421, "expires_at": "2026-08-25T18:00:00+00:00"}

▸ Rotate key

$ curl -X POST https://phantom.codes/v1/key/rotate \
    -H "Authorization: Bearer YOUR_PHANTOM_KEY"

{"api_key": "sk-...", "balance_usd": 9.7421}

Old key dies instantly. Plaintext returned exactly once. Store it.

// 03

Streaming

Pass "stream": true. SSE-compatible. Each chunk arrives as data: {...} followed by a blank line. Final chunk before data: [DONE] carries the usage block used for billing.

$ curl -N https://phantom.codes/v1/chat/completions \
    -H "Authorization: Bearer YOUR_PHANTOM_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "phantom/qwen3.5-122b-a10b",
      "stream": true,
      "messages": [{"role":"user","content":"Hi"}]
    }'

data: {"id":"chatcmpl-...","choices":[{"delta":{"content":"He"}}]}

data: {"id":"chatcmpl-...","choices":[{"delta":{"content":"llo"}}]}

data: {"id":"chatcmpl-...","choices":[{"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":8,"completion_tokens":2,"total_tokens":10}}

data: [DONE]

Abort mid-stream: phantom bills for tokens already generated. If upstream sent no bytes, no charge. Intentional aborts before max_tokens finishes are still billed.

// 04

Image generation

OpenAI-compatible. Flat-rate per image. Upstream URLs live ~1 hour. Download them.

$ curl https://phantom.codes/v1/images/generations \
    -H "Authorization: Bearer YOUR_PHANTOM_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "stability/stable-diffusion-3-5-large",
      "prompt": "A misty cyberpunk alley at dawn",
      "n": 1,
      "size": "1024x1024",
      "quality": "standard"
    }'

Parameters: model, prompt (max 4000 chars), n (1-10), size (256/512/1024/1536/2048 square), quality (standard | hd), response_format (url | b64_json). Per-model pricing in /v1/models under image_pricing_usd_user.

// 05

Attestation

Bind each response to verified TEE hardware in three steps.

1. Run any chat completion. Capture response.id.
2. GET /v1/signature/{id}?model=... → returns signature + signing_address.
3. GET /v1/inference-attest?model=...&nonce=$(openssl rand -hex 32)&signing_address=... → returns Intel TDX quote + NVIDIA CC payload.

Verify offline:

• TDX quote → POST https://phantom.codes/v1/verify/tdx
• NVIDIA payload → POST https://phantom.codes/v1/verify/gpu
• Signature → recover pubkey, confirm matches signing_address

Chain: response → signature → TEE quote.

// 06

Integrations

Dedicated coding-agents page with full setups: /code.html.

Anything that takes base_url + api_key works. Set:

$ export OPENAI_BASE_URL=https://phantom.codes/v1
$ export OPENAI_API_KEY=sk-your-phantom-key

Picks up automatically in: LangChain, LlamaIndex, Vercel AI SDK, Pydantic-AI, llm, mods, aichat, shell-gpt, tgpt, chatblade.

▸ Tool-by-tool config

Aider:

aider --openai-api-base https://phantom.codes/v1 \
      --openai-api-key  sk-your-phantom-key \
      --model           openai/phantom/deepseek-v4-flash

OpenHands:

export LLM_API_KEY=sk-your-phantom-key
export LLM_BASE_URL=https://phantom.codes/v1
export LLM_MODEL=openai/phantom/deepseek-v4-flash

opencode · ~/.config/opencode/opencode.json:

{
  "provider": {
    "phantom": {
      "npm": "@ai-sdk/openai-compatible",
      "options": {
        "baseURL": "https://phantom.codes/v1",
        "apiKey": "{env:OPENAI_API_KEY}"
      },
      "models": {
        "phantom/deepseek-v4-flash": {"name": "DeepSeek V4 Flash"}
      }
    }
  }
}

Crush · ./crush.json:

{
  "$schema": "https://charm.land/crush.json",
  "providers": {
    "phantom": {
      "type": "openai-compat",
      "base_url": "https://phantom.codes/v1",
      "api_key": "$OPENAI_API_KEY",
      "models": [
        {
          "id": "phantom/deepseek-v4-flash",
          "name": "DeepSeek V4 Flash",
          "context_window": 1048576,
          "default_max_tokens": 8192
        }
      ]
    }
  }
}

type must be "openai-compat" — "openai" is reserved for OpenAI itself. Leave models empty and Crush pulls the whole catalog from /v1/models.

Continue.dev · ~/.continue/config.json:

{
  "models": [{
    "title":    "Phantom DeepSeek",
    "provider": "openai",
    "model":    "phantom/deepseek-v4-flash",
    "apiKey":   "sk-your-phantom-key",
    "apiBase":  "https://phantom.codes/v1"
  }]
}

Cline / Cursor / OpenWebUI: pick "OpenAI Compatible" in settings. Base URL = https://phantom.codes/v1, API Key = your sk-..., Model = any from /v1/models.

{
  "model": "phantom/deepseek-v4-flash",
  "messages": [{"role": "user", "content": "Weather in Tokyo?"}],
  "tools": [{
    "type": "function",
    "function": {
      "name": "get_weather",
      "description": "Current weather for a city",
      "parameters": {
        "type": "object",
        "properties": {"city": {"type": "string"}},
        "required": ["city"]
      }
    }
  }],
  "tool_choice": "auto"
}

// 07

Rate limits

Hashed in memory. Never logged, never persisted.

// 08

FAQ