š³ Monthly Suscription Pricing#
Pay $500āÆUSD / 500K credits ā build and scale fast
š Pricing Components at a Glance#
| Component | What You Pay For | Rate / Unit | Billing Frequency |
|---|
| Infrastructure | Compute time while your job is running on our clusters | per second | Include in your suscription |
| LLM Tokens | Combined inputāÆ+āÆoutput tokens processed by the model | provider list price Ć tokens | On completion of each job |
| Storage | Persisted docs & vector embeddings leveraged for retrieval | provider list price x size | It is usually monthly |
| ShadAI API | Calls that trigger agent actions | 500KāÆcredits / month | Include in your suscription |
š Detailed Breakdown#
1. Infrastructure#
Every time an agent runs, it spins up secure, autoāscaling compute.
2. LLM Tokens#
Agents call bestāināclass LLMs. Youāre charged for both directions:Input tokens (your prompt + system context)
Output tokens (modelās response)
We simply pass through each providerās published token price so you can choose the model that fits your budget/performance profile.
3. Storage#
Documents and vectors ingested during sessions fuel semantic search. You can configure your own Postgres database.
4. ShadAI API Usage#
Using the ShadAI Client directly or via agents. Each call (sync or async) counts toward your quota.$500āÆUSD ā 500K credits monthly
Recharge in USD credits#
In billing page you can recharge credits to use ShadAI. Only you need to press ADD CREDITS button and follow next steps. The minimum amount allowed is 500 USD.
š” Tips to Optimize Cost#
1.
Choose the right model ā use smaller/faster models for highāvolume tasks.
2.
Keep sessions alive when possible.
3.
Don't ingest multiple times same data to reduce storage.
š°Ā LLM Token Pricing CheatāSheet (MayĀ 2025)#
All prices are USD per token. Multiply byĀ 1āÆ000 to estimate cost perāÆK. Always verify with vendor docs before production use.
šµ GoogleĀ Ā·Ā Gemini
| Model | InĀ ($) | OutĀ ($) |
|---|
| GEMINI_2_0_FLASH | 0.00000070 | 0.00000040 |
| GEMINI_2_0_FLASH_LITE | 0.00000008 | 0.00000008 |
| GEMINI_1_5_FLASH | 0.00000015 | 0.00000060 |
| GEMINI_1_5_PRO | 0.00000250 | 0.00000100 |
| GEMINI_1_5_FLASH_8B | 0.00000008 | 0.00000030 |
š§ AnthropicĀ Ā·Ā ClaudeĀ 3
| Model | InĀ ($) | OutĀ ($) |
|---|
| CLAUDE_3_7_SONNET | 0.00000300 | 0.00001500 |
| CLAUDE_3_5_SONNET_2024_10_22 | 0.00000300 | 0.00001500 |
| CLAUDE_3_5_SONNET | 0.00000300 | 0.00001500 |
| CLAUDE_3_5_HAIKU | 0.00000100 | 0.00000400 |
| CLAUDE_3_OPUS | 0.00001500 | 0.00007500 |
| CLAUDE_3_SONNET | 0.00000300 | 0.00001500 |
| CLAUDE_3_HAIKU | 0.00000030 | 0.00000400 |
š§® DeepSeekĀ Ā·Ā RĀ series
| Model | InĀ ($) | OutĀ ($) |
|---|
| DEEPSEEK_R1_V1 | 0.00000135 | 0.00000540 |
š¦ MetaĀ Ā·Ā Llama
| Model | InĀ ($) | OutĀ ($) |
|---|
| LLAMA_4_MAVERICK_17B_INSTRUCT | 0.00000024 | 0.00000097 |
| LLAMA_4_SCOUT_17B_INSTRUCT | 0.00000017 | 0.00000066 |
| LLAMA_3_3_70B_INSTRUCT | 0.00000070 | 0.00000072 |
| LLAMA_3_2_1B_INSTRUCT | 0.00000010 | 0.00000010 |
| LLAMA_3_1_70B_INSTRUCT | 0.00000070 | 0.00000072 |
| LLAMA_3_1_8B_INSTRUCT | 0.00000020 | 0.00000022 |
š CohereĀ Ā·Ā Command
| Model | InĀ ($) | OutĀ ($) |
|---|
| COHERE_COMMAND_PLUS | 0.00000300 | 0.00001500 |
| COHERE_COMMAND_R | 0.00000050 | 0.00000150 |
| COHERE_COMMAND_LIGHT | 0.00000030 | 0.00000060 |
āļø AmazonĀ Ā·Ā Nova
| Model | InĀ ($) | OutĀ ($) |
|---|
| AMAZON_NOVA_PRO | 0.00000080 | 0.00000320 |
| AMAZON_NOVA_LITE | 0.00000006 | 0.00000024 |
| AMAZON_NOVA_MICRO | 0.00000004 | 0.00000014 |
š”Ā Tips#
Use Flash Lite / Haiku for prototypes, then scale up as accuracy demands.
Double-check whether prices include input and output tokens separately (some dashboards only report combined cost).
Remember storage & inference infra are billed separately.
Modified atĀ 2025-09-30 05:04:50
