Live Feed
PerceptionBench: No Frontier Model Breaks 60% on Visual Perception — GPT-5.6 Sol Leads at 59.7%
Moonshot AI's new benchmark isolates visual perception from reasoning across 10 atomic sub-skills. The best frontier model scores 59.7%. The weakest category is object hallucination, where GPT-5.6 Sol manages 26.9% and a weaker Gemini 3.5 Flash beats it at 50.6%.
Google's $15B India AI Hub Takes Shape in Andhra Pradesh — Largest Overseas AI Build on Earth
Google's $15B data centre hub in Andhra Pradesh, breaking ground in April 2026, is the company's largest AI investment anywhere outside the United States. Three sites, 601 acres, TPU and GPU compute, and subsea cable infrastructure.
Riot Platforms Exits Bitcoin Mining With $9.8B in AI Contracts — Anthropic Deal Closes the Pivot
Anthropic signed a $9.1B, 20-year lease with Riot Platforms for 191 megawatts at Rockdale, Texas. Combined with a January AMD deal, Riot now holds $9.8B in contracted AI revenue and has effectively converted from Bitcoin miner to AI data center landlord.
Anthropic's August 2026 Risk Report Names Two Unreleased Frontier Models — All Claude Lines Provisionally Hit CB-1
Anthropic's biannual safety assessment discloses two internal frontier-class models that were not publicly released as of July 15, including one broadly comparable to Mythos 5. Every Claude line has now provisionally crossed the CB-1 catastrophic-harm threshold since Opus 4.
Debian Puts AI Policy to an 8-Way Vote — From Social Contract Ban to Climate Refusal
The Debian project opened voting today on 8 competing AI/LLM policy proposals, ranging from a total Social Contract ban to a climate-based rejection. Voting runs August 15 to 28. This is the project's first binding decision on AI contributions after two earlier attempts failed.
GLM-5.3 Tops CyberGym at 84.5%, Finds a Cursor Vulnerability, Ships Open Weights in Two Weeks
Z.ai's GLM-5.3 posts 66.9% on DeepSWE (+20 points over GLM-5.2), leads CyberGym at 84.5% ahead of GPT-5.6 Sol and Mythos 5, and reportedly found a serious vulnerability in Cursor before public release. Open weights follow once safety hardening is done.
Arena Adds Gemini 3.7 Flash High and Grok 4.6 High Across Three Leaderboards in 48 Hours
Six models entered Arena across multiple leaderboards between August 11 and 13. Gemini 3.7 Flash High now has head-to-head preference data in Agent, Text, and Code arenas. Grok 4.6 High — carrying 95.6% SWE-bench — is live in Code and Text.
SK Hynix Locks Korea's HBM Capacity Through 2033 — and the US Has No Domestic Wafer Fab
SK Hynix's board vote commits its AI memory roadmap to Korea through 2033. The US has zero HBM wafer fabrication capacity, with its earliest realistic domestic timeline arriving in 2028 — a packaging facility only, not wafer production. Every NVIDIA GPU and Google TPU runs on Korean-made HBM.
Grok 4.6 Lands in GitHub Copilot — SpaceXAI's 95.6% SWE-Bench Model Reaches Every VS Code Developer
SpaceXAI pushes Grok 4.6 into Copilot's model picker at the same $2/$6 per million pricing as Grok 4.5, but with a fundamentally different model underneath: 95.6% SWE-bench Verified, Intelligence Index 61, and a new entry on Arena's Code and Text leaderboards.
Databricks Closes $5B at $190B — Up 42% in Six Months, Coatue Leads
Databricks has closed a $5 billion strategic round at a $190 billion valuation, up from $134 billion in February 2026 and $62 billion in January 2025. Capital targets Lakebase, the Genie AI coworker, and Unity AI Gateway.
DeepSeek V4-Pro 0813 Hits 96.4% on Neutral SWE-Bench — Open Weights at #2, 59x Cheaper Than Claude Opus 5
Vals.ai's neutral bash harness puts DeepSeek V4-Pro-0813 at 96.4% on SWE-bench Verified, second overall and ahead of GPT-5.6 Sol. The MIT-licensed open model costs $0.022 per task against Claude Opus 5's $1.29.
Jeff Dean's Discovery Loop in Talks for $1B at $10B to Automate Science
Discovery Loop, the public benefit corporation Jeff Dean founded after 27 years at Google, is raising $1 billion at a $10 billion valuation. The startup's stated aim is to automate machine learning, science, and technology.
Terminal-Bench 3.0 Ships: Frontier Agents Drop From 84% to 34% as the Bar Resets
Scale Labs and the Harbor community ship Terminal-Bench 3.0 on July 23, replacing tasks frontier models were approaching 80% on with a harder problem set. GPT-5.6 Sol Max and Claude Fable 5 Max now lead at 34% — the new ceiling — while Grok 4.6 reaches 26% and Grok 4.5 falls to 15.7%.
DeepSeek V4 Pro and Flash Pricing Jumps Up to 12x on August 16 as Peak/Off-Peak Rates Replace Flat Billing
DeepSeek restructures API pricing around peak and off-peak windows effective August 16, with V4-Pro cache-hit input rates rising more than 12x at peak. The change coincides with V4-Pro's official GA and lands while the model scores just one point above V4-Flash on the Artificial Analysis Intelligence Index.
Google Audits 75 AI Research Papers and Finds Evidence Failures in Every System Tested
Google Cloud AI Research ran five autonomous research agents across 75 papers on five tasks and found every baseline had at least one systematic evidence failure — fabricated citations, scores that did not reproduce, or method claims that were not in the submitted code. ScientistOne proposes a Chain-of-Evidence framework to close the gap.