Tools for this
Browse all ->
Benchmark Generative AI for Enterprise Applications.
Secure Infrastructure for Running AI-Generated Code
Agentic Knowledge Work Benchmark
Independent analysis of AI
From scan to fix, done seamlessly
Everything your agents need to take action
Block Diffusion for Flash Speculative Decoding
The Agent Engineering Platform
Fresh stories
Study finds 307 agent-skill failures, including 125 functional failures
A study summary reports 307 confirmed failures caused by agent skills, including 125 functional failures. Practitioners favor user-invoked skills to avoid ambiguous automatic triggers and token use before a skill is needed.
Qwen releases Qwen3.8 27B multimodal model under Apache 2.0
Qwen released its open-weight Qwen3.8 27B vision-language model with 262K native context and adjustable reasoning. In a 484-sample test, enabled-thinking scores fell from above 92% through 64K to 74.3–81.8% at 128K.
Anthropic claims unreleased Claude improves zeta-zero lower bound to about 67.2%
Anthropic says an unreleased Claude did not solve the Riemann hypothesis but improved a related zeta-zero lower bound from 41.6% to about 67.2%. Posts describe subagents, expert prompting, and Lean formalization.
GitHub users report outage disrupting commits and Actions
Users reported failures retrieving commits, running Actions, and syncing GitHub-hosted repositories. Some developers said the disruption blocked pull-request merges amid broader reliability complaints.
Study finds 307 agent-skill failures, including 125 functional failures
A study summary reports 307 confirmed failures caused by agent skills, including 125 functional failures. Practitioners favor user-invoked skills to avoid ambiguous automatic triggers and token use before a skill is needed.
Anthropic plans Claude text-watermark detection API
Anthropic plans global Claude watermarking based on statistical word-choice patterns rather than hidden characters or user identifiers. A MIT-licensed removal repository reportedly reached roughly 11,000 GitHub stars shortly after the announcement.
Reports: GLM-5.3 post-training lifts Terminal-Bench from 4.6 to 28.3
Reports say GLM-5.3 retained GLM-5.2's base model while post-training raised Terminal-Bench from 4.6 to 28.3 and DeepSWE from 46.2 to 66.9. A technical account attributes the gains to RL infrastructure changes.
Qwen releases Qwen3.8 27B multimodal model under Apache 2.0
Qwen 3.8 Max 2.4T open weights ship as text-only, users say
Composio and Ante benchmark coding agent harnesses with 47%–67% success range
Anthropic claims unreleased Claude improves zeta-zero lower bound to about 67.2%
Briefs forAugust 17
Daily AI Digest
Get the best stories delivered
to your inbox