Proprietary inference stack at 10,000 tokens/sec for Warp #8858
Description
Pre-submit Checks
- I have searched Warp feature requests and there are no duplicates
- I have searched Warp docs and my feature is not there
Describe the solution you'd like?
Hi,
We built a proprietary stack at Kog combining a bare metal inference engine in Assembly/C++ and a parallel model architecture that generates entire sequences simultaneously rather than token by token, reaching 10,000 tokens per second per request where standard providers sit around 100 t/s.
We are looking for teams running generation-heavy workflows like complex multi-repo code changes executed by parallel agents to validate this together.
What would that concretely change for the Warp team?
Is your feature request related to a problem? Please describe.
No response
Additional context
No response
Operating system (OS)
Select an OS
How important is this feature to you?
4
Warp Internal (ignore) - linear-label:39cc6478-1249-4ee7-950b-c428edfeecd1
None