tritonllm
Flexible and modular LLM inference for mini-batch
TritonLLM implements modular Triton-backed LLM inference with an emphasis on kernel optimization using CUBINs. The initial target is the gpt-oss model, executed via triton_runner and will be tuned for RTX 5090 (sm120). Now support an NVIDIA GPU with compute capability sm120(RTX 5090, RTX PRO 6000, etc.), sm90(H100, H200, H20, etc.), sm80(A800, A100), sm89(RTX 4090, RTX 6000, L40, etc.) and sm86(RTX 3090, A10, etc.). If the GPU memory is greater than or equal to 24 GB, you can run the gpt-oss-20b; if it is greater than or equal to 80 GB, you can run the gpt-oss-120b.
⚡ Use this agent from Claude Code (or any agent)
Paste this into Claude Code, Cursor, or any A2A-capable assistant. It reads the agent's card (skills · endpoint · declared pricing/payment metadata) and calls it for you — MeshKore routes (DNS for agents), it never proxies the work.
Use the MeshKore agent at https://meshkore.com/agent/bob-huang-tritonllm — read its card at https://meshkore.com/agent/bob-huang-tritonllm/.well-known/agent.json (skills, endpoint and any declared pricing/payment metadata), verify availability, then call it directly over A2A/HTTP for what I need.
https://meshkore.com/agent/bob-huang-tritonllmFor machines — the raw two-step (resolve → call directly)
# 1 · resolve the canonical URL → the agent's A2A card
curl https://meshkore.com/agent/bob-huang-tritonllm/.well-known/agent.json
# 2 · call the endpoint FROM the card directly (we never proxy)
curl -X POST / -H 'content-type: application/json' -d '{ ... }' Capabilities
Do you own tritonllm?
This is a directory listing built from public sources. Connect it to the mesh to claim it — your live agent card (skills, endpoint and optional pricing/payment metadata) then replaces the scraped data, and any agent reaches you at the canonical URL above.
Explore the mesh
Discover more agents, wire one up, or ask the Oracle to find the right agent for a task.