Full archive

Benchmark: Latest AI News

Search reporting on AI markets, models, products, companies, policy, and culture.

Showing 1-6 of 6Results matching "data centers"

Earlier this week

1 story

Sep 1, 2026

Grok 4.6 Clears 50% on Biosecurity Refusal and Routine Biology Tasks

The reported result rewards a difficult balance: recognizing hazards hidden inside ordinary-looking research requests without shutting down legitimate biological work. Its surveillance score shows that refusal performance is not the whole biosecurity test. The result puts the model’s ability to distinguish intent at the center of a biosecurity contest where excessive blocking can also impair useful work.

3 min readRead story

Last week

4 stories

Benchmark Business

Aug 25, 2026

Amazon Puts $25B Standalone Marker on Chip Business While Remaining a Top Nvidia Customer

Jassy’s hypothetical revenue benchmark does not describe a separately reported company, but it gives Amazon’s internal chip effort a new measure of scale as it both buys from and competes with Nvidia. Amazon CEO Andy Jassy has put a new number on the company’s semiconductor ambitions: its chip operation would generate more than $25 billion in annual revenue if it stood alone.

2 min readRead story

Benchmark Infrastructure

Aug 25, 2026

OpenAI’s Jalapeño Claims Up to 1.9x Efficiency Gain, but Leaves Nvidia Training Intact

The Broadcom co-developed inference processor is headed for deployment by year-end. Its published gains are meaningful, but they exclude model training and Nvidia’s Vera Rubin platform. OpenAI says Jalapeño, its custom chip co-developed with Broadcom, delivered up to 1.9 times higher throughput per kilowatt and 1.7 to 3.6 times lower end-to-end latency than Nvidia GB200 and GB300 systems in published tests.

3 min readRead story

Benchmark Products

Aug 25, 2026

Tiangong Ultra Runs 100m in 8.86 Seconds, Then Hits a Stopping Mat

The Beijing race shows a sharp jump in humanoid sprint performance, but its finish also exposed a basic operational constraint: Tiangong Ultra needed a thick mat to stop and caught fire after impact. Tiangong Ultra, a humanoid robot from the Beijing Humanoid Robot Innovation Center, ran 100 metres in 8.86 seconds in a large-size division semifinal at Beijing’s World Humanoid Robot Games. High jump: 2.88 metres.

3 min readRead story

Benchmark Models

Aug 25, 2026

HiDream-O1-World Tops WBench Navi at 80.9, Betting on Persistent 3D Scenes

HiDream.ai’s new model is designed to make generated environments hold together through movement and edits. Its strongest disclosed evidence is a navigation benchmark; its larger film, robotics and production ambitions remain proposed uses. HiDream.ai has released HiDream-O1-World, a model intended to turn text prompts, images, or interactive controls into explorable 3D worlds.

3 min readRead story

August 2026

1 story

Benchmark Tools

Aug 23, 2026

InferenceX Adds 11 Telemetry Views to AgentX, Exposing What Benchmark Curves Hide

The new exploration layer makes cache setup, bursty subagents and queue behavior inspectable, but its best-curve design means readers must still examine each point’s underlying configuration before treating a curve as a like-for-like comparison. A throughput curve can make two inference systems look nearly identical. The distinction reflects the workload AgentX is designed to measure.

3 min readRead story