Meet Tera Compute, from the team behind Hashbranch

Hi there,

You signed up with Hashbranch, so we wanted to share what we’ve been building.

The team behind Hashbranch just launched Tera Compute (tera.gw), focused on AI inference, the compute that runs AI models.

Tera serves open-weight models on GPUs we run in the US. Open-weight models are the open-source AI models, like Llama, DeepSeek, Qwen, Kimi and GLM, that anyone can run, as opposed to closed models like GPT or Claude that you can only rent from one company. They’re what most teams now use to build AI agents and apps.

Why builders use Tera:

• Cheaper: a fraction of closed-model pricing, output from $0.25 per million tokens.

• Private: US-hosted, zero data retention. Your prompts are never stored or trained on.

• No lock-in: every model in our open catalog on one API key, OpenAI-compatible, switch with one line.

If you or your team are building anything with AI, we want to put $150 of free credits in your hands to try it. No card, no contract.

Fill a 2-minute form and we’ll email your key with the credits on it. No call required.

Hashbranch is still running. Tera is what we’re building next on the AI side, and as an early Hashbranch member you’re among the first to see it.

Deepak and Tom

Founders, Tera Compute