#Pricing

Pay only for what you use. No subscriptions, no minimums. All prices are in USD per 1M tokens.

##Chat completions

Model	Input	Output	Cached input*	Reasoning**
`gob-5.5-scout`	$0.15	$0.60	$0.075	$0.30
`gob-5.5`	$2.50	$10.00	$1.25	$5.00
`gob-5.5-deep`	$3.50	$14.00	$1.75	$7.00
`gob-5.5-horde`	$5.00	$15.00	$2.50	$7.50

\ Cached input applies when the same prefix is reused within 5 minutes (50% discount).* \\* Reasoning tokens — used by Goblin-of-Thought when reasoning: "visible" is set — are billed at 50% of the output rate.*

##Embeddings

Model	Price per 1M tokens
`gob-embed-cave-small`	$0.02
`gob-embed-cave`	$0.10
`gob-embed-cave-large`	$0.13
`gob-embed-shadow`	$0.18

##Goblin Reward Signal API

Endpoint	Price per 1M input tokens
`/grs/score`	$0.05

Output is a single scalar — no output token charge.

##Cave Memory

Charge	Price
Hoard storage	$0.10 / 1M hoards / month
Hoard retrieval (per call)	included with chat call (no separate charge)
Hoard extraction (per session)	included with last chat call

##Per-request cost calculator

Approximate cost of a typical chat completion:

Scenario	Tokens (in/out)	Model	Cost
Short Q&A	50 / 200	scout	$0.000128
Standard chat reply	500 / 800	gob-5.5	$0.00925
Code generation	2k / 4k	gob-5.5	$0.045
Multi-step research	10k / 8k	gob-5.5-deep	$0.147
Agentic workflow with cave memory	30k / 12k	gob-5.5-horde	$0.330

##Free tier

gob-test- keys are completely free with these caps:

Resource	Cap
Chat completions	10,000 / month
Embeddings	1M tokens / month
GRS scoring	100k tokens / month
Cave Memory storage	1k hoards

Test-key responses are functionally identical to production but may be slightly delayed during peak hours.

##Volume discounts

Auto-applied based on monthly spend:

Monthly spend	Discount
$0 – $1,000	0%
$1k – $10k	5%
$10k – $50k	10%
$50k – $250k	15%
$250k+	contact sales (up to 30%)

##Prepaid credits

Buy credits in advance for additional discounts:

Prepaid amount	Bonus
$100	5%
$500	8%
$2,000	12%
$10,000	18%

Credits never expire.

##Billing

▸Billed in arrears at the end of each calendar month.
▸Itemized invoice with per-model and per-endpoint breakdown.
▸Auto-charged to your payment method on the 1st.
▸Failed payments suspend the API after 7 days. No data is deleted.

##Cost monitoring

Set hard and soft spend alerts in Console → Billing. When you cross 80% of your soft alert, you get an email. When you cross your hard cap, the API starts returning no_treasure_in_hoard (429).

You can also poll usage programmatically:

bash

curl https://api.gpt-gob.ai/v1/usage?start_date=2026-05-01&end_date=2026-05-31 \
  -H "Authorization: Bearer $GOB_API_KEY"

##What's included free

▸All API key management
▸Cave Memory infrastructure (storage charges separate)
▸All SDKs and tooling
▸Standard support (community Discord + email)
▸Status page + incident notifications

##What costs extra

▸Dedicated capacity (enterprise)
▸VPC/private deployment
▸24/7 phone support
▸Custom fine-tuning
▸On-prem deployment