πŸ—‚οΈ Subnet Services Directory

Frontend services offered by each Bittensor subnet

← Tracker

πŸ”₯ Confirmed & Tested

Gravity / Macrocosmos SDK
SN13 β€” Data Universe
βœ… Tested

What: Batch + on-demand data API (X/Twitter, Reddit). SDK with GravityClient, Sn13Client, BillingClient.

Pricing: $0.05/1k rows (batch) Β· $0.10/1k rows (on-demand) Β· Credits don't expire Β· TAO payment coming soonβ„’

Free tier: $5.00 credits

Best for: Research, guide content, data exploration. Not suitable as direct miner feed β€” validators detect duplicate data.

macrocosmos.ai β†’
Desearch / Datura
SN22 β€” Already staking (19.82Ο‡)
βœ… Active

What: Decentralized web search API. We use it for research and SN70 Vericore miner integration.

Open question: Can we pay in TAO instead of USD?

desearch.ai β†’
Chutes
SN64 β€” Rayon Labs Β· Decentralized AI Inference
βœ… Active

What: Serverless AI inference β€” deploy and run open-source models. #1 subnet by usage. 160B+ tokens/day. Deploys new HuggingFace models within 24h of release.

How we use it: Integrated directly into OpenClaw as a model provider. We run DeepSeek V3, Mistral, GLM and others via Chutes aliases for cost-efficient tasks.

Pricing: ~$0.22/M tokens β€” roughly 85% cheaper than AWS. All revenue goes to alpha token buybacks.

Notable: TEE support for private inference. Serves majority of free DeepSeek on OpenRouter. #3 DePIN project by revenue (~$6M annual).

Models: DeepSeek V3, Mistral Small, GLM-4, Qwen, Kimi, MiniMax and more β€” updated constantly.

chutes.ai β†’

πŸ”¬ Researched β€” Not Yet Tested

Targon
SN4 β€” Manifold Labs Β· GPU Cloud

NVIDIA H200s & 4090s, confidential compute. Usage-based pricing. GPU-focused β€” unlikely fit for general hosting.

targon.com β†’
Sybil
Manifold Labs Β· AI Search

Open-source model inference via Bittensor. Could partially replace LLM costs for utility tasks. No Claude/GPT.

sybil.com β†’
Nineteen AI
SN19 Β· Text & Image Inference

Behind Vercel checkpoint β€” needs manual browser visit to check pricing. Potential LLM inference alternative.

nineteen.ai β†’
Corcel
Bittensor AI Inference

Site was unreachable as of May 2026. May be deprecated or migrating.

corcel.io β†’

πŸ“Š Realistic Migration Assessment

Expense Est. Monthly Best Alternative TAO Payment?
Apify (scraping) ~$65 cap Gravity (research only) Soonβ„’
LLM Inference Variable Sybil (open-source only) Unknown
Hosting ~$52/top-up Targon (GPU-focused, unlikely) Unknown
Proxy (TPN) ~$10 No ecosystem equivalent N/A
LLM Inference (Chutes) ~$0.22/M tokens Already using βœ… Alpha token
Desearch API ~$10 Already ecosystem βœ… Ask Datura
Last updated: May 22, 2026 Β· ← Back to Tracker