AI Coding Live News

On November 4, Tencent Hy published Hy4 on its official site. Versus Augusts Hy4 preview, Tencent says it scaled up pre-training and post-training, added vision for the first time, added computer use across CLI and GUI, and trained specifically on Blender, CAD and engines like Unreal, Unity and Godot. Hy4 is live on Tencent Cloud and OpenRouter (tencent/hy4, 1M context) at the same API price as the preview. Weights and a system card are promised soon, and a cheaper Hy4 flash is coming.

Vals benchmarked Tencent Hy4 Preview at #21 of 58 on the Vals Index (55.4%) and #4 among open-weight models at about $1.28 per test. It is the cheapest top-ten model on Code Migration (47.4%) versus much costlier Muse Spark and Opus 4.8 runs, leads open-weight models on three agentic benches, and ships with a 1M context window and 64k max output tokens.

Arena.ai (@arena) reports Agent Arena results: DeepSeek-V4.1-Flash (Max) posts about +4.87% net improvement at roughly $0.07 median task cost — lowest among the top-3 open models — retaining ~98% of Hy4 preview’s net gain at ~73% lower cost and ~76% of Kimi K3 (Max) performance at ~92% lower cost.

Official @ggml_org shipped llama.cpp v0.4.1: adds Maple 20B-A1B, Tencent Hy 4, and Spark2.5 support; improves Kimi-K3 rollback, JSON schema handling, qwen3-coder complex-type parsing, and MCP/multimodal fixes; updates ggml to v0.24.0. Release notes: ggml-org/llama.cpp v0.4.1.

vLLM shipped v0.29.0 with Model Runner V2 as the default runner for all models (594 commits, 277 contributors). Highlights include another Kimi-K3 optimization pass, DeepSeek-V4 shared experts fused into MegaMoE, Mamba prefix caching that keeps internal prefill checkpoints (about 9–25% better TTFT), per-request speculative-decoding acceptance stats over the API, a sharded_rdt P2P backend for RL weight sync, and new model support such as Hy4-preview, Qwen3.8-Flash-Next, GraniteSWA, and NemotronH Omni Reasoning V3.

Cline said Tencent Hy4 Preview currently leads SWE-bench Pro. The open-weight model is described as 770B parameters with 49B active and a 1M-token context window, and Cline called it Tencent's biggest generational leap measured to date. The practical developer path is immediate: npm i -g cline, run /model, and select Hy4 preview. SWE-bench Pro is a harder, more realistic software-engineering eval than classic HumanEval-style completion, so a lead there is a stronger signal for repository-level agents than a short coding quiz. The 1M context window is the other half of the story: long-horizon refactors, multi-file patches, and eval traces can stay in a single session instead of being retrieved in fragments. Cline's distribution also matters; putting Hy4 beside GLM-5.3 and Qwen3.8 in the same agent makes model shopping a /model command rather than a new IDE.