Which AI writes the better take? You decide — blind.

Two top models go head-to-head on today's AI news. Pick the sharper summary without seeing the names — the crowd's verdict builds the leaderboard.

Agents & InferenceHacker News

ZCode 3.0 optimizes GLM-5.2 for multi-agent collaboration

Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.

Summary A

ZCode 3.0 is a GLM-5.2-optimized coding-agent harness with multi-agent task orchestration across planning, code changes, review, and release automation. For teams shipping agentic dev workflows, the practical value is less raw model access and more an opinionated workspace/task layer around GLM-5.2, which can speed evaluation but may duplicate or constrain existing internal agent infrastructure.

What you'll learn · Jul 2, 2026 · 6 stories

  1. 1.GLM-5.2 gains faster multi-agent workflows in ZCode 3.0; teams can now plan, code, and deploy with lower coordination overhead.
  2. 2.GLM-5.2 integration in ZCode 3.0 speeds up multi-agent coding workflows for planning, review, and deployment without toolchain changes.
  3. 3.31B Gemma 4 on Cerebras hardware reduces P95 latency to under 300ms, enabling real-time voice agents in 9,000+ robots.
  4. 4.Sonnet 5 cuts inference costs by ~20% vs Opus 4.8 while matching accuracy, but English tokens cost 1.4x more than Sonnet 3.5.
  5. 5.Default blocks on ad-heavy pages may cut AI training data by half unless crawlers split search from agent use.
  6. 6.Gemini Spark beta on Mac lets Ultra subscribers automate file tasks and app workflows, competing with Claude and Copilot at no added cost.
Browse editions · 43 days
Agents & InferenceHacker News

ZCode 3.0 adds multi-agent support for GLM-5.2 coding tasks

Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.

Summary A

ZCode is positioning GLM-5.2 as a Claude Code-style coding agent stack with 20+ programming-tool integrations, selected MCP tools, multi-agent collaboration, and remote task start via WeChat, Feishu, or Telegram. For teams shipping agentic dev workflows, this means GLM is now a serious alternative backend to evaluate at the workflow layer—repo access, tool permissions, review/release automation, and reproducibility matter more than raw model quality alone.

Agents & InferenceHugging Face

Hugging Face and Cerebras bring Gemma 4 to real-time voice AI

Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.

Summary A

Gemma 4 31B is now shown in an open speech-to-speech loop on Cerebras, with Parakeet ASR and Qwen3TTS, targeting low and stable latency rather than just better median response time. For production voice agents and robots, the key implication is that the LLM step can stop being the long-tail bottleneck, making natural turn-taking feasible in modular open stacks already deployed across 9,000+ Reachy Mini robots.

Agents & InferenceSimon Willison

What's new in Claude Sonnet 5

Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.

Summary A

Sonnet 5 delivers near-Opus-4.8 quality at lower per-token prices, but ships a new tokenizer that produces ~1.4x more tokens for English, ~1.33x for Spanish, and ~1.28x for Python—so your effective cost per document rises even as the headline rate drops. Re-benchmark your actual token counts before assuming a cost win, and don't reuse old token-based budget or context-window estimates; Mandarin workloads are the exception where costs stay flat.

Agents & InferenceTechCrunch

Cloudflare’s new policy pushes AI companies to pay for publishers’ content

Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.

Summary A

Starting September 15, 2026, Cloudflare will block mixed-use AI/search crawlers by default on ad-supported pages for new sites, new customers, and existing free customers unless owners opt out. If you rely on web crawling for training, retrieval, or agents, you need separate, transparent crawler identities for search versus AI use and should expect more publisher content to move behind permissioned or paid access rather than being freely fetchable.

Agents & InferenceTechCrunch

Gemini Spark, Google’s agentic assistant, is now available on Mac

Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.

Summary A

Gemini Spark now supports custom MCP servers on macOS, meaning you can wire your own tools and apps directly into Google's desktop agent alongside its native file access and new first-party integrations (Keep, Tasks, Canva, Dropbox, Instacart). Practically, this puts Spark in the same MCP-based tooling ecosystem as Claude Desktop, so your existing MCP integrations become portable across agents—but it's gated behind AI Ultra, US-only beta, with remote/multi-step task orchestration still not shipped.

See who's winning the model face-off

Tomorrow's blind matchup and the running leaderboard — one email a day.

Takeaways written by Mistral Large — not one of this week's two contestants.