Long Horizon Terminal Benchmark with Dense Reward Grading
-
Updated
Oct 12, 2026 - Python
Long Horizon Terminal Benchmark with Dense Reward Grading
Tencent Hy3 295B MoE (NVFP4) on 2x NVIDIA DGX Spark — TP2 over 200GbE, 256K context, 26 tok/s end-to-end. Tuned, benchmarked, agent-ready.
Hyprland window flipping, with experimental multi-app cards, unfold, peek, saved layouts and configurable transitions.
Hunyuan HY3-powered proactive learning agent that continuously tracks long-horizon learning goals, verifies real progress, and actively intervenes with reminders, reviews, teaching, and plan adjustments.
First public recipe for Tencent Hy3-295B (295B MoE, NVFP4) across two NVIDIA DGX Spark GB10 nodes: vLLM TP=2, MTP speculative decoding, and the full toolchain-debugging story including what fails.
HY3 Desktop — Tencent Hunyuan 3 assistant that also codes, runs agents & builds Excel/decks. One-click, free 1 month. Apache 2.0. Win/Mac.
Read-only Hy3 spec-to-diff reviewer with a loopback Web UI and CLI, local citation validation, and Markdown/JSON PR-readiness reports via TokenHub.
Hy3-powered microscopy, FLIM, and scientific RAG research agent
Issue-killer: Hy3-powered GitHub Issue contribution planner for Tencent Rhino-Bird 2026
Run real Claude Code through CLIProxyAPI and WorkBuddy HY3 with local tool calling
Hy3-powered local context relay for AI coding tools
A multi-provider AI coding gateway with Anthropic-compatible APIs, MCP integration, and orchestration support for models from Anthropic, DeepSeek, MiMo, MiniMax, Kimi, Z.ai, and OpenRouter.
To associate your repository with the hy3 topic, visit your repo's landing page and select "manage topics."