Sign in

Official Endpoints · Wholesale Prices

DeepSeek, Kimi, GLM & Qwen Official Endpoints, Wholesale Prices.

Every request is served by the model maker's own API. One key, USD billing, 20% below official list. No regional signup friction.

Get An API Key See The Prices
−20% vs official listOfficial upstreamPay as you goNo credit card needed
Launch PricingUSD / 1M
Kimi K3Output / 1M Tokens
$15.00
$13.50Save 10%
DeepSeek V4-ProOutput / 1M Tokens
$1.98
$1.58Save 20%
Qwen3.8 MaxOutput / 1M Tokens
$6.00
$4.80Save 20%
DeepSeek V4.1-Flash$0.60$0.48−20%DeepSeek V4-Pro$1.98$1.58−20%GLM-5.3 Flash$0.50$0.40−20%GLM-5.3$4.40$3.52−20%GLM-4.7$2.20$1.76−20%Kimi K3$15.00$13.50−10%Kimi K2.7-Code$4.00$3.20−20%Qwen3.8 Max$6.00$4.80−20%Qwen3.7 Plus$1.60$1.28−20%

Output Prices · USD / 1M Tokens

Pricing

The same official endpoints.
But at a lower cost.

Nine models. Transparent input and output pricing. Save 20% against official prices on every model below, except Kimi K3 at 10%.

Launch pricing · October 12, 2026USD / 1M tokens · input → output
ModelOfficialOpenRouter(official endpoint)BulkTokenYou save
DeepSeek V4.1-Flash$0.15 → $0.60$0.16 → $0.63$0.12 → $0.48−20%
DeepSeek V4-Pro$0.66 → $1.98$0.70 → $2.09$0.53 → $1.58−20%
GLM-5.3 Flash$0.15 → $0.50$0.16 → $0.53$0.12 → $0.40−20%
GLM-5.3$1.40 → $4.40$1.48 → $4.64$1.12 → $3.52−20%
GLM-4.7$0.60 → $2.20$0.63 → $2.32$0.48 → $1.76−20%
Kimi K3$3.00 → $15.00$3.17 → $15.83$2.70 → $13.50−10%
Kimi K2.7-Code$0.95 → $4.00$1.00 → $4.22$0.76 → $3.20−20%
Qwen3.8 Max$2.00 → $6.00$2.11 → $6.33$1.60 → $4.80−20%
Qwen3.7 Plus$0.40 → $1.60$0.42 → $1.69$0.32 → $1.28−20%

DeepSeek shows off-peak rates; peak rates are twice as high. Peak hours: Monday–Friday, 01:00–04:00 and 06:00–10:00 UTC, excluding Chinese public holidays. Qwen uses International list prices; Qwen3.7 Plus shows the tier up to 256K input tokens. OpenRouter figures include a 5.5% credit purchase fee. Displayed rates are rounded to two decimal places. See the model catalog for availability and full billing details.

Run your own numbers.

Choose a model and your monthly spend at official prices to estimate your savings.

$50 / mo$50,000 / mo

Official spend: $1,000/mo · BulkToken: $800/mo

You save · 20%$200/mothat's $2,400 per yearEstimate based on the selected model’s discount.

Fair question

Why are our prices lower?

Cheap API access has a trust problem, and it should. So here is exactly where our discount comes from — and where it doesn't.

01

We buy in bulk

Tier-1 volume agreements with the model providers. We pre-purchase token capacity at wholesale rates — the same economics as any wholesale business. Bulk in, bulk out.

02

Thin margin, high volume

We keep a small spread and pass the rest of the discount to you. Our pricing ceiling is the official list — never the floor we build markup on top of.

03

No subscription games

Pay as you go, credits never expire, no tiers gating the models. We earn when you build — not when you forget a recurring charge.

What we're not

Low prices usually mean someone else is paying the bill. Not here. Wholesale pricing is sustainable — that's the whole point.

  • Not account pools or stolen credentials
  • Not resold consumer subscriptions
  • Not free-trial or promo-credit arbitrage
  • Not third-party-hosted lookalike weights
  • Every request goes to the model maker's own API. If that ever changes for any model, it will say so on the pricing page — in writing.

Verify it yourself

Don't trust us.
Test us.

Every claim on this page is checkable in under five minutes. We'll even pay for the test.

  • Same prompt, both endpoints

    Run identical prompts against the official API and BulkToken. Compare outputs and logprobs yourself — same weights, same behavior.

  • Full feature parity

    Streaming, tool calling, JSON mode, prompt caching — every official capability passes through untouched.

  • Full-precision official weights

    Never quantized, distilled, or swapped for a cheaper model. If a model can't meet that bar, we don't list it.

  • Live status, in public

    Per-model uptime and latency on our status page — no signup required.

Claim $2 Free Credits
verify.py
from openai import OpenAI
import os

prompt = "Explain how prompt caching works."

for name, url, key in [
    ("official", os.environ["OFFICIAL_BASE_URL"],
     os.environ["OFFICIAL_API_KEY"]),
    ("bulktoken", "https://bulktoken.ai/v1",
     os.environ["BULKTOKEN_API_KEY"]),
]:
    client = OpenAI(base_url=url, api_key=key)
    response = client.chat.completions.create(
        model=os.environ["MODEL_ID"],
        messages=[{"role": "user", "content": prompt}],
    )
    print(name, response.choices[0].message.content)
    print(response.usage)

One API

One API for all models

DeepSeek, GLM, Kimi and Qwen behind a single OpenAI-compatible endpoint. Switch models by changing one string.

DeepSeek V4.1-Flash

$0.60$0.48 out−20%

DeepSeek V4-Pro

$1.98$1.58 out−20%

GLM-5.3 Flash

$0.50$0.40 out−20%

GLM-5.3

$4.40$3.52 out−20%

GLM-4.7

$2.20$1.76 out−20%

Kimi K3

$15.00$13.50 out−10%

Kimi K2.7-Code

$4.00$3.20 out−20%

Qwen3.8 Max

$6.00$4.80 out−20%

Qwen3.7 Plus

$1.60$1.28 out−20%

Same SDK. Same workflow.
A lower bill.

  • No Chinese phone number
  • No regional identity checks
  • No RMB-only payment rails
  • No per-model accounts
  • USD billing — card, USDT & USDC
  • Credits never expire
migrate.py
from openai import OpenAI
import os

client = OpenAI(
    api_key=os.environ["BULKTOKEN_API_KEY"],
    base_url="https://bulktoken.ai/v1",
)

resp = client.chat.completions.create(
    model="qwen3.8-max",
    messages=[
        {"role": "user", "content": "Hello!"}
    ],
)

print(resp.choices[0].message.content)

Quick Start

Up and running in minutes.

  1. 01

    Get your API key

    Create an account and generate a key in the console. One key works for every model.

  2. 02

    Add Credits

    Prepay in USD. Use your Credits across every model.

    Add Credits
  3. 03

    Make your first request

    Use your API key with an OpenAI-compatible SDK and send your first request. https://bulktoken.ai/v1

Set up with AI

Copy the integration guide and paste it into Cursor, Claude Code or any coding assistant.

Read the Docs
first_request.py
from openai import OpenAI
import os

client = OpenAI(
    api_key=os.environ["BULKTOKEN_API_KEY"],
    base_url="https://bulktoken.ai/v1",
)

resp = client.chat.completions.create(
    model="qwen3.8-max",
    messages=[
        {"role": "user", "content": "Hello!"}
    ],
)

print(resp.choices[0].message.content)

Volume & Enterprise

More Volume.
Lower Rates.

Building at scale? Get deeper discounts, tailored API support, and company invoicing for your growing workload.

Deeper Volume Discounts

Get custom rates based on the models you use and your expected volume.

Dedicated API Support

Work with us on API integration, capacity requirements, and ongoing technical support.

Company Support

Get help with company invoicing, account management, and payment arrangements for your business.

Let's find your best rate.

Tell us which models you need and your expected monthly usage. We'll put together a pricing and support proposal for your workload.

Contact Sales
Prefer to send your requirements?
Get a Custom Quote

For high-volume developers, teams, and businesses.

FAQ

Questions, answered.

Why are your prices lower?

We procure model capacity at wholesale rates through volume agreements, and pass the difference on to you. There are no subscriptions or seat fees — you simply pay per call at a lower listed rate.

Which models are available?

The launch lineup includes DeepSeek V4.1-Flash, DeepSeek V4-Pro, GLM-5.3 Flash, GLM-5.3, GLM-4.7, Kimi K3, Kimi K2.7-Code, Qwen3.8 Max and Qwen3.7 Plus.

Does every model have the same discount?

Kimi K3 is 10% below its official price. All eight other launch models are 20% below their official prices. Input and output rates are shown separately.

How does billing work?

Add Credits in USD, then pay as you go: each request deducts Credits based on actual usage and the model’s listed rates. Language models bill input and output tokens separately, with cache pricing where supported. Other models use the billing unit shown on their model page.

Can I use the same Credits across different models?

Yes. Your Credits are shared across the models and API keys on your account. You do not need to top up separately for each model or key.

How do I migrate an existing project?

Use the BulkToken base URL and API key with your OpenAI-compatible client, then choose a model ID from the Models page. Our documentation includes runnable examples.

I have high volume usage. Can I get custom pricing?

If you are a developer or enterprise with large token consumption, reach out to us for a custom quote. Email: [email protected]

Official endpoints. Wholesale prices.
Start testing and building now

$2 of free credits. No card required. Five minutes is all it takes to verify.

Claim $2 Free Credits