Nate@nate_512

Seguir

SubQ Snap long-context model launch

Free output tokens is the part I keep coming back to. SubQ launched Snap, and the input side runs $0.05 per million tokens, which is a number I had to read twice. Snap handles up to 384K tokens of multimodal context, and SubQ reported sub-second decisions on 100K-token inputs. That combo is what gets me. Long context plus that speed plus basically no cost on the way out. Document-heavy workflows suddenly look a lot less painful to mess around with. First 10,000 users also get 100M free tokens 👀

1d

Nate@nate_512

Seguir

SubQ Snap long-context model launch

Free output tokens is the part I keep coming back to. SubQ launched Snap, and the input side runs $0.05 per million tokens, which is a number I had to read twice. Snap handles up to 384K tokens of multimodal context, and SubQ reported sub-second decisions on 100K-token inputs. That combo is what gets me. Long context plus that speed plus basically no cost on the way out. Document-heavy workflows suddenly look a lot less painful to mess around with. First 10,000 users also get 100M free tokens 👀

1d