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

Show HN: OpenKnowledge – open source AI-first alternative to Obsidian/Notion

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

Summary A

OpenKnowledge has been introduced as an open-source, AI-first alternative to popular productivity and organization tools like Obsidian and Notion. The platform focuses on integrating artificial intelligence capabilities directly into daily workflows. It offers users a customizable option for managing and structuring their information.

What you'll learn · Jun 26, 2026 · 6 stories

  1. 1.Show HN: OpenKnowledge – open source AI-first alternative to Obsidian/Notion
  2. 2.45°C cooling could reduce data center water demand to near zero, useful where water use constrains AI infrastructure expansion.
  3. 3.75% growth since January suggests Claude is broadening beyond developers, but ChatGPT still has many more paying users.
  4. 4.15-fold revenue growth signals strong demand for simulated environments that catch agent shortcuts before deployment in software engineering and finance.
  5. 5.AI agents can take on longer, more complex tasks, expanding productivity across roles where multi-step work slows teams down.
  6. 6.One command gives teams a temporary OpenAI-compatible endpoint for tests, evals, or batch generation, billed by hardware usage instead of managed production service.
Browse editions · 43 days
Agents & InferenceHacker News

45°C cooling design cuts data center water use to near zero

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

Summary A

An innovative cooling design operating at 45°C is capable of reducing data center water usage to near-zero levels. This advancement represents a more sustainable approach to thermal management in large-scale computing infrastructure.

Agents & InferenceTechCrunch

Anthropic’s Claude is winning over paid consumers, a market owned by ChatGPT

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

Summary A

Anthropic's AI assistant Claude is rapidly gaining popularity among paying consumers, a market segment long dominated by OpenAI's ChatGPT. Credit card transaction data shows Claude's paying user base and revenue have grown by approximately 75% since January 2026, alongside a massive surge in consumer search interest and educational course demand. While ChatGPT remains the overall market leader by a wide margin, Claude is steadily securing a larger share of consumer spending and interest.

Agents & InferenceTechCrunch

Patronus AI lands $50M to build ‘digital worlds’ that stress-test AI agents

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

Summary A

Patronus AI has raised $50 million in Series B funding to develop "digital worlds" that test and fine-tune AI agents performing complex tasks. The startup's simulated environments help model makers and companies evaluate AI performance across various scenarios. Patronus' revenue has grown 15-fold over the past year, driven by demand from frontier AI labs and emerging startups.

Agents & InferenceOpenAI

How agents are transforming work

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

Summary A

AI agents are transforming the workplace by enabling employees to handle longer and more complex tasks. A new OpenAI research paper highlights the expansion of productivity across various roles due to AI agents. This development is changing the nature of work.

Agents & InferenceHugging Face

Run a vLLM Server on HF Jobs in One Command

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

Summary A

Hugging Face has streamlined the deployment of private, OpenAI-compatible large language model endpoints using a single command on its HF Jobs infrastructure. The process operates on a pay-per-minute hardware billing system, allowing users to spin up a vLLM server without provisioning servers or managing Kubernetes. Once active, the secure endpoint can be queried from any location using a Hugging Face API token.

See who's winning the model face-off

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

Takeaways written by GPT-5.5 — not one of this week's two contestants.