70% off for AI models.

One key, one endpoint — lower prices across leading AI providers.

Start saving
  • No platform fees
  • No routing fees
  • Pay as you go
  • OpenAI
  • Anthropic
  • Google
  • DeepSeek
  • Qwen
  • Kimi
  • z.ai
  • xAI
  • AionLabs
  • MiniMax
  • Tencent
  • Xiaomi
  • Meta
  • ByteDance
Live catalog

Model Pricing

ModelDiscountInput / 1MOutput / 1M24h volumeScore
AnthropicLong contextReasoning
40% off
$6.00Official $10.00
$30.00Official $50.00
24h volume118B
Score81.5
AnthropicLong contextReasoning
40% off
$6.00Official $10.00
$30.00Official $50.00
24h volume104.8B
Score92.4
AnthropicReasoning
No discount
$4.00
$20.00
24h volume88.3B
Score97.6
Qwen
30% off
$0.07Official $0.10
$0.28Official $0.40
24h volume87.4B
ScoreScore unavailable
Z.aiReasoning
45% off
$0.77Official $1.40
$2.42Official $4.40
24h volume69.9B
Score73.1
GoogleLong contextReasoning
20% off
$1.60Official $2.00
$9.60Official $12.00
24h volume66.3B
Score77.0
AnthropicLong contextReasoning
40% off
$3.00Official $5.00
$15.00Official $25.00
24h volume65.1B
Score81.9
MoonshotReasoning
35% off
$0.6175Official $0.95
$2.60Official $4.00
24h volume64B
Score74.7
QwenReasoning
30% off
$1.75Official $2.50
$5.25Official $7.50
24h volume52.3B
Score74.7
Showing 9 of 62 matching models

Transparent usage

See where every dollar goes.

Total spend$804.25
Example usageLast 7 days
Daily spend
Spend by model

Guaranteed cache hits.

Fall below your harness’s guaranteed rate in a month, and you get Tokun usage credit.

  • 95%

    • Prime Agent
    • Pi
    • Grok Build
    • Codex
    • Claude Code
    • OpenCode
    • OpenClaw
    • Hermes

No Fees

$0 platform fees. $0 routing fees. Only pay for the tokens you use.

OpenAI-compatibleSwap the base URL and the key. Existing SDKs and clients keep working.
Streaming & toolsToken streaming, tool calls and structured output pass through untouched.
Automatic failoverA degraded provider is skipped mid-route instead of returned as an error.
Per-key limitsIssue keys per service and cap spend without opening a support ticket.
Zero prompt retentionRequests are proxied, not stored. Tokun logs metadata for billing — never prompt or completion bodies.
Transparent billingEvery line item shows the model, the route taken and the price paid.
Usage exportPull the same numbers the dashboard shows straight into your own reporting.
No lock-inStandard model names in, standard responses out. Leave whenever you want.

Sell capacity

Tell us where you have available capacity and its approximate dollar value. Our team will review the fit and contact you directly.

  1. Share your capacity
  2. Get reviewed
  3. Start serving requests

Spend less. Build more.

Lower model prices. Guaranteed cache hits. One API.

Start saving
  • No platform fees
  • No routing fees
  • Pay as you go