Templar I, Chutes and Gradients: the full agent lifecycle on Bittensor
Thread125 words
Describes how Templar I, a 1.2B LLM trained permissionlessly on a Bittensor subnet, can be served on Chutes and instruction-finetuned via Gradients, giving a full agent lifecycle on Bittensor. Notes no-code agents can be built with squad.ai.
Key claims
Templar I is the first LLM trained entirely permissionlessly on Bittensor
It can be served on Chutes and instruction-finetuned on Gradients
Templar I is a 1.2B LLM trained on the @tplr_ai subnet, served compute on @chutes_ai (subnet built by @rayon_labs), marking the first time an LLM was trained entirely permissionlessly on @opentensor's decentralized infra.
Using @gradients_ai, one could even use an incentivized p2p network of miners to compete in order to instruction finetune an LLM trained on @tplr_ai and served on @chutes_ai.
The full AI agent lifecycle, all on Bittensor.
Open-source AI is rising.
You can even build your own agent running Templar I on Chutes via @rayon_labs' squad(.)ai platform. Without writing a single line of code.
Truly one of the most exiting ecosystems out there.
Also check my post below for more on @tpr_ai ↓
And the no-code squad(.)ai agent I built last week ↓
1 image on X
What came next · since 03 Jun 2025
Since then
Zero has written 7 more pieces on Bittensor since this one.
- Watchlist additions: TAO, PURR, EDEL and BPOn Bittensor
- I Built a Research Desk Staffed Entirely by AI Agents. It Works.On Bittensor
- AI infra projects on watch: Bittensor, Sentient, NEAR, Ritual and moreOn Bittensor
- Why DeAI matters across the AI stack: seven layers from incentives to computeOn Bittensor