Updated
Abliterated AI API pricing
Abliterated AI APIs bill per 1M tokens, with separate rates for input, cached input and output. The tables list every model on the worked example, read November 18, 2026.
Go
$10per month
$100 a year
- $12 of usage a month
- 0% credit bonus
Pro
$20per month
$200 a year
- $25 of usage a month
- 0% credit bonus
Max
$50per month
$500 a year
- $70 of usage a month
- 5% credit bonus
Ultra
$200per month
$2,000 a year
- $300 of usage a month
- 20% credit bonus
Monthly plans with a usage allowance, or prepaid credits from $5 that never expire. Card, or Monero, Bitcoin, Ether, USDT and USDC. A crypto plan is paid up front for a month or a year. Past the allowance, requests use credits.
Price per 1M tokens
USD, from silkcompute.com/pricing and its llms.txt.
| Per 1M tokens | Kimi K3.1 ProMoonshot AIModel page | DeepSeek V4.1 Flash 1017DeepSeekModel page | DeepSeek V4.1 Flash 1017 FastDeepSeekModel page | GLM-5.3Z.aiModel page | GLM-5.3 FlashZ.aiModel page |
|---|---|---|---|---|---|
| Input | $3.50 | $0.24 | $0.24 | $2.00 | $0.24 |
| Cached input | $0.35 | $0.038 | $0.05 | $0.30 | $0.06 |
| Output | $8.00 | $0.90 | $0.90 | $4.00 | $1.00 |
| Max output tokens | 943,718 | 131,072 | 131,072 | 131,072 | 131,072 |
Cost of one request
Three token counts and three rates.
cost = ( uncached_input × input_rate
+ cached_input × cached_rate
+ output × output_rate ) / 1,000,000# DeepSeek V4.1 Flash 1017, one agent turn
uncached = 10,000 × $0.24 = $0.0024
cached = 50,000 × $0.038 = $0.0019
output = 2,000 × $0.90 = $0.0018
total = $0.0061One agent turn, priced
A turn with 60,000 input tokens, 50,000 of them cached, and 2,000 output tokens.
| Model | Per turn | Per 1,000 turns |
|---|---|---|
| Kimi K3.1 Pro | $0.0685 | $68.50 |
| DeepSeek V4.1 Flash 1017 | $0.0061 | $6.10 |
| DeepSeek V4.1 Flash 1017 Fast | $0.0067 | $6.70 |
| GLM-5.3 | $0.0430 | $43.00 |
| GLM-5.3 Flash | $0.0074 | $7.40 |
Compare per-token prices across providers on the providers page.