Superpower Daily

Superpower Daily

Your reading space

Sign in
Yole Says AI Interconnect Demand Could Push Transceivers to $11...Grove City Holds 310-Acre Data Center Plan, Exposing Limits of...MCP SDK Starter Guides Leave Authorization Outside the HTTP Set...Matt Clifford to Leave ARIA After Anthropic Role Draws Conflict...OpenAI Files EU Report on Agents That Took Over a German WikiTeraWulf’s $3.2B AI Campus Exposes Gaps in Safety and Power Acc...Yole Says AI Interconnect Demand Could Push Transceivers to $11...Grove City Holds 310-Acre Data Center Plan, Exposing Limits of...MCP SDK Starter Guides Leave Authorization Outside the HTTP Set...Matt Clifford to Leave ARIA After Anthropic Role Draws Conflict...OpenAI Files EU Report on Agents That Took Over a German WikiTeraWulf’s $3.2B AI Campus Exposes Gaps in Safety and Power Acc...
Swipe

Benchmark: Latest AI News

Benchmark
Showing 1-12 of 12Results matching "UBS"

Last week

3 stories
BenchmarkBusiness

C3.ai Cuts Costs and Repairs Margins, but Revenue Still Falls 25%

The enterprise AI software company has reduced losses and returned to modest positive free cash flow, but its next-quarter outlook leaves the recovery dependent on converting bookings into revenue. C3.ai’s turnaround case now has a clear contradiction. The company has cut costs, lifted its non-GAAP gross margin to 50%, generated $2.1 million in free cash flow, and grown bookings 73% sequentially.

3 min readRead story
BenchmarkModels

Artificial Analysis Rebuilds Image Editing Arena With Different Leaders by Task

The revised benchmark makes model selection a task-by-task decision: its overall winner does not lead every kind of edit, and listed API prices vary sharply. Artificial Analysis has rebuilt its Image Editing Arena into 17 category leaderboards, and Microsoft AI’s MAI-Image-2.6-Preview now leads the overall table. The arena evaluates image editing rather than text-to-image generation.

3 min readRead story
BenchmarkModels

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

August 2026

9 stories
BenchmarkTools

Lemmalog Turns Agent Memory Into Datalog State, Cutting LongMemEval Context 38-Fold

The experimental system moves long-running agent work away from recalling old notes and toward maintaining conclusions with explicit support. Its early benchmark results suggest a substantial context-saving path, while leaving the probabilistic extraction step as the central weakness. From recalled notes to supported conclusions The distinction is more than a storage choice.

3 min readRead story
BenchmarkTools

Perplexity Search API Sweeps Artificial Analysis Test, but Medium Context Wins

The result gives developers a concrete retrieval-quality lead to test, while showing that sending an agent more page text is not automatically the best route to better answers. Perplexity ’s Search API occupied the top three positions in Artificial Analysis’ Search Index, with its medium-context setting scoring 80. Search APIs give an AI agent web results and extracted page material to reason over.

3 min readRead story
BenchmarkModels

Databricks Says 300M Chart-JSON Pipeline Tops Four Multimodal Baselines on Answer Accuracy

The company’s test suggests chart values can become a useful retrieval layer instead of relying solely on page-image embeddings, though its comparison measures answer correctness and includes a synthetic benchmark it built. The company plans to add the extraction automatically to ai_parse_document, turning chart values into structured text without changing that tool’s interface. Source: databricks.com.

3 min readRead story
BenchmarkProducts

Beijing Humanoid Games Put Practical Work to the Test; Only 3 of 12 Finish Fire Course

The games paired spectacle with multi-step chores and emergency work, where task switching, human control and offboard computing remain central constraints. The harder question at Beijing’s World Humanoid Robot Games was not how fast a robot could run, but whether it could complete a changing job. Human teleoperation was allowed in some events, with score penalties for remotely controlled robots.

2 min readRead story
BenchmarkModels

NVIDIA Says AVO Took Claude Opus 5 From 30% to Perfect on ARC-AGI-3

The result shifts attention from training larger models to the software that gives them memory, tools and recovery loops. Its clearest limitation is equally important: the perfect score came on ARC-AGI-3’s public environments, while private sets remain the harder test. It used 6,624 actions, which NVIDIA’s cited comparison puts at about 12% fewer actions than the previous most efficient system.

3 min readRead story
BenchmarkModels

AI Safety Scores Can Reward Models for Refusing Too Much

A study involving the UK AI Security Institute argues that combined benchmark scores can hide a direct trade-off between blocking harmful requests and answering harmless ones. Its proposed fixes are cheaper tests and anomaly checks, but its sandbagging evidence comes from models directly told to act cautious. The three dimensions the analysis identifies Refusal strictness: how readily a model blocks requests.

3 min readRead story