Auto-Arch Tournament vs GLM-5V-Turbo

Auto-Arch Tournament is an autonomous research system where an AI agent iteratively proposes, implements, and validates microarchitectural improvements to a RISC-V CPU. Starting from a standard 5-stage pipeline, the loop runs hypotheses in parallel, each going through formal verification (53 symbolic checks), cycle-accurate simulation, multi-seed FPGA place-and-route, and CoreMark CRC validation. Only hypotheses that beat the current champion get merged; everything else gets discarded. Starting from 301 iterations/second, the system hit 577 iter/s (+92%) across 73 attempts in 9.8 hours — producing a design 26% faster and 40% smaller in LUTs than the baseline. The insight the author drives home is that the real innovation isn't the AI agent — it's the verifier. The orchestrator is hardcoded to prevent agents from manipulating their own evaluation gates, a simple but critical design constraint that turns a creative process into a trustworthy one. Without a rigorous verification harness, agent-driven optimization becomes a confidence trick. This is early but fascinating proof that AI-driven hardware design loops can produce commercially meaningful gains. The repo uses Claude Code or Codex as the coding agent, SystemVerilog for the RTL, and standard open-source EDA tooling (Yosys, nextpnr, Verilator). It's a compelling template for anyone building agentic optimization loops where correctness matters.

GLM-5V-Turbo is a multimodal vision-language model from Zhipu AI (international brand: Z.ai) purpose-built for converting visual designs into executable code. Released April 3, 2026, it's optimized specifically for the design-to-code pipeline that's becoming central to AI-assisted frontend development. The model features a 200K token context window with 128K max output — enough to hold an entire design system plus generate substantial implementation code in a single call. Input support spans images, video, and text. The CogViT vision encoder was trained from scratch alongside the language model rather than bolted on post-training, which Zhipu claims is why it achieves 94.8 on the Design2Code benchmark vs. Claude Opus 4.6's 77.3 (their own testing). GUI agent workflows are a first-class use case, with strong results on AndroidWorld and WebVoyager benchmarks. Pricing is competitive at $1.20/M input tokens and $4/M output tokens, with free web access at chat.z.ai for exploration. For teams already doing design-to-code workflows with Figma exports and Claude, GLM-5V-Turbo is a direct challenger worth benchmarking — especially given the claimed 17-point lead on the primary evaluation.