Skip to content
tokenblackhole.
Menu

A public ledger of token prices

Everything you spend on tokens disappears. Pick the cheapest hole.

Every row below is ranked by one number: how many dollars of DeepSeek-equivalent usage you get back for each dollar you pay. Promotions, subscriptions, cache-hit rates and the fine print included.

Best right now
20x
B.AI
Deals tracked
9
each with a source and a date
Measured against
1x
DeepSeek's own list price

The ledger

Ranked by what a dollar buys.

A 20x row is not automatically better than a 7x row. Read the caveats — expiry and rate caps decide the real answer.

  1. B.AI

    Read caveatsSource: mixed

    b.ai/verified 2026-09-20

    20x if you top up with BNB, 15x with anything else

    Covers

    DeepSeek V4.1 Flash

    $10.00 → $200.00

    Maximum multiplier

    20x

    Details
  2. Volcano Engine

    Read caveatsSubscriptionSource: unconfirmedEstimated

    www.volcengine.com/verified 2026-09-20

    ¥49 tier worth ¥200 — 4x on paper, maybe 4x again on top

    Covers

    Doubao and other hosted models

    $1.00 → $4.00

    Maximum multiplier

    ~16x

    Details
  3. Alibaba Cloud

    Read caveatsSubscriptionSource: unconfirmed

    www.alibabacloud.com/verified 2026-09-20

    TokenPlan: Lite 4.5x, Standard 6x, Pro 6.3x — double it off-peak

    Covers

    Qwen and other hosted models

    $1.00 → $4.50

    Maximum multiplier

    12.6x

    Details
  4. InferX

    Read caveatsSource: mixed

    inferx.net/verified 2026-09-20

    10x on selected models — but the context stops at 262k

    Covers

    200+ models; most endpoints capped at 262,000 context

    $1.00 → $10.00

    Maximum multiplier

    10x

    Details
  5. Command Code

    VerifiedSubscriptionSource: primary

    commandcode.ai/verified 2026-09-20

    Go plan: the cheapest way in, at 10x

    Covers

    DeepSeek V4.1 Flash, GLM, Qwen and more

    $1.00 → $10.00

    Maximum multiplier

    10x

    Details
  6. Command Code

    VerifiedSubscriptionSource: primary

    commandcode.ai/verified 2026-09-20

    GOAT plan: $10 a month for $70 of model credits

    Covers

    DeepSeek V4.1 Flash, GLM, Qwen, Kimi, GPT-5.6, 30+ models

    $10.00 → $70.00

    Maximum multiplier

    7x

    Details
  7. OpenCode

    VerifiedSubscriptionSource: unconfirmed

    opencode.ai/verified 2026-09-20

    Go plan: 6x on the same open models

    Covers

    DeepSeek V4.1 Flash and other open models

    $1.00 → $6.00

    Maximum multiplier

    6x

    Details
  8. LLM Gateway

    Read caveatsSource: unconfirmed

    llmgateway.io/verified 2026-09-20

    3x today, 2x shortly — a rate that is on its way down

    Covers

    Routed across many providers

    $1.00 → $3.00

    Maximum multiplier

    3x

    Details
  9. DeepSeek

    VerifiedSource: primary

    api.deepseek.com/verified 2026-09-20

    Official pay-as-you-go — the baseline every deal is measured against

    Covers

    DeepSeek V4.1 Flash, DeepSeek V4 Pro

    $10.00 → $10.00

    Maximum multiplier

    1x

    Details

About this site

This site looks like AI slop because an AI built it. Every plan on it is one I actually pay for and use.

@aimaster3658on Threads

What does your burn rate actually cost?

Priced at DeepSeek's published off-peak rates for V4.1 Flash, with your cache-hit rate applied. The multiplier is then applied on top.

Accepts 10B, 1.5B, 250M, or a plain number.

An agent re-sending its context runs 95%+. A one-shot chat runs near 0%.

Monthly at list price

$8,803.80

Monthly on the cheapest route

$440.19

via B.AI at up to 20x — you keep $8,363.61 a month

Every route, same burn

  • B.AI$440.1920x
  • Volcano Engine$550.2416x
  • Alibaba Cloud$698.7112.6x
  • InferX$880.3810x
  • Command Code$880.3810x
  • Command Code$1,257.697x
  • OpenCode$1,467.306x
  • LLM Gateway$2,934.603x

Priced at 10B tokens/day over 30 days.

Read this before you top up

The fine print is the product.

Multipliers are the headline. Expiry, rate caps, per-model allowances and counterparty risk are what actually decide your bill.

Ask in the chat