Public beta · drop-in OpenAI replacement

Switch OpenAI to
Chinese frontier models,
cut AI spend 90%.

DeepSeek v4-flash, Kimi K2.6, GPT-5.5, Claude Opus 4.7, Gemini 2.5 Pro —— 30+ models, one API, zero hassle. No Chinese entity required. Pay in USD.

one line change, 100× cheaper

from openai import OpenAI

client = OpenAI(
    base_url="https://promptoll.com/v1",
    api_key="zai_•••",
)

client.chat.completions.create(
    model="deepseek-v3.1",   # or "auto" — we pick the cheapest
    messages=[{"role": "user", "content": "Hello"}],
)

30+ models, one API — no lock-in

Official model weights, no distillation. Frontier + Chinese value models side by side.

🌐 Western Frontier

◉OpenAI

gpt-5.5Flagship

gpt-5.4

gpt-5.4-miniBest value

gpt-5.4-nano

gpt-5.3-codexCode

o3Reasoning

o3-miniReasoning

o4-miniReasoning

gpt-4o

gpt-4.1

gpt-4.1-mini

gpt-4o-mini

gpt-4.1-nano

🧠Anthropic

claude-opus-4-7Flagship

claude-opus-4-6

claude-opus-4-5-20251101

claude-sonnet-4-6Recommended

claude-sonnet-4-5-20250929

claude-haiku-4-5-20251001

🔮Google

gemini-2.5-pro-thinkingReasoning

gemini-2.5-proRecommended

gemini-2.5-flash-thinkingReasoning

gemini-2.5-flash

⚡xAI

grok-4.3FlagshipLive

🔍Perplexity

sonarWeb search

🇨🇳 Chinese Value

🐋DeepSeek

deepseek-v4-proFlagship

deepseek-v4-flashRecommendedFast

deepseek-v3.1

🌙Moonshot Kimi

kimi-k2.6Flagship

kimi-k2.5

🎵 Media models · Embeddings / ASR / TTS / Image / Video

text-embedding-3-smalltext-embedding-3-largetext-embedding-ada-002whisper-1tts-1tts-1-hddall-e-3dall-e-3-hdsoraveo-2vidu-2.0vidu-q1-t2v

Why this actually works

Chinese labs competed harder on price-performance. Your bill is the beneficiary.

🔌

Drop-in OpenAI replacement

Same SDK, same message format, same streaming semantics. Change one line and you're done.

♻

Smart arbitrage routing

Send model="auto" and we route to the cheapest model that meets your quality bar. Opt in per-tenant.

📊

Token-accurate billing

Every request logs prompt/completion tokens at microcent precision. See cost, revenue, margin per request.

🔑

Per-key model binding

Issue scoped keys locked to a specific model. Sell fixed-price SKUs to your own customers.

🌍

No Chinese entity required

⚡

Global edge delivery

Routes via the lowest-latency path for your users — US, EU, APAC all covered.

Live in under 5 minutes

No contracts. No minimums. Point back to OpenAI any time.

Sign up

Email + password. Start using immediately after registration.

Mint an API key

Console → API Keys → Create. Optionally bind it to a specific model.

Swap base_url

Point your OpenAI SDK at promptoll.com/v1. Code stays the same.

Watch your bill drop

Every request shows cost / price / model used in the dashboard.

Transparent pricing

Same as calling the provider direct.

Prices match each provider's official rate card — no markup, no hidden fees. You get access without the card / VPN friction.

Official price per 1K input tokens

model             $ / 1K input tokens
─────────────────────────────────────
gpt-4o            $0.00250
claude-sonnet-4-6 $0.00300
gpt-4.1-mini      $0.00040
─────────────────────────────────────
deepseek-v3.1     $0.00027   ←  9× vs gpt-4o
deepseek-v4-flash $0.00014   ← 18× vs gpt-4o
kimi-k2.5         $0.00060
gemini-2.5-flash  $0.00030

Ship AI without the wallet bleed.

Top up and start calling — pay as you go, balance never expires.

Switch OpenAI to Chinese frontier models,cut AI spend 90%.