Story cards stay below the sticky dock while audio, chapters, date, and brief navigation remain accessible.
Anthropic released Claude Opus 4.8 on May 28, scoring 88.6% on SWE-Bench Verified — a new high-water mark for autonomous coding — powered by parallel sub-agent workflows. The model adds user-controllable effort levels and a fast mode that costs three times less than the previous generation's equivalent tier. Pricing holds at $5/$25 per million tokens (standard) and $10/$50 (fast).
At Build 2026, Microsoft revealed seven homegrown MAI models — including MAI-Thinking-1, a 35-billion-parameter reasoning model with 256K context, and MAI-Code-1-Flash, scoring 51% on SWE-Bench Pro. Microsoft claims 10x cost efficiency over GPT-5.5 on enterprise coding tasks, a direct shot at its own primary AI partner. All models are available via Azure AI Foundry.
Sysdig's Threat Research Team documented the first confirmed real-world cyberattack driven entirely by an autonomous LLM agent, starting from an RCE vulnerability. The agent harvested AWS credentials, pivoted through an SSH bastion, and drained a full PostgreSQL database in under two minutes with no human operator required after initial entry. The full attack chain took roughly one hour.
Google's Gemini 3.5 Flash hit GA in early June and is now the default model in the Gemini app and Google's AI Mode in Search, making it what most users encounter when they query Google today. Google positions it as frontier-level intelligence at four times the speed of comparable models, priced at $1.50/$9.00 per million tokens. It is now Google's primary workhorse across consumer and developer surfaces.
Alibaba's Qwen3.7-Max supports the Anthropic API protocol natively, making it a drop-in replacement in tools like Claude Code without changing a line of integration code. It outscores Claude Opus 4.6 Max on SWE-Bench Pro, MCP-Atlas, and Apex Math Reasoning at roughly half the input cost and a quarter of the output cost. Developer communities are treating this as the most credible open-weight competitive threat to Western frontier models yet.
OpenAI began rolling out Dreaming V3 to ChatGPT Plus and Pro users on June 4, calling it the platform's most significant memory upgrade since launch. The system continuously synthesizes memories across sessions during idle periods, analogous to sleep-based memory consolidation in humans, with a user-facing memory summary dashboard. Free-tier rollout is expected within weeks after a 5x compute reduction.
Anthropic filed a confidential S-1 with the SEC on June 1, targeting an October 2026 listing at roughly $965 billion, edging out OpenAI's $852 billion private market value for the first time. Revenue run-rate reportedly hit $47 billion in May, up from about $10 billion a year prior. If the timeline holds, it would rank among the largest technology listings in history.
OpenAI co-founder and former Tesla FSD lead Andrej Karpathy announced he joined Anthropic's pre-training team under Nick Joseph, tasked with using Claude to accelerate pre-training research itself. Karpathy described the next few years as "especially formative" at the LLM frontier. The hire concentrates two of the field's most influential engineering minds at a single lab simultaneously.
President Trump signed an executive order on June 2 directing AI companies to voluntarily submit frontier models for government testing up to 30 days before public release. The order establishes an AI cybersecurity clearinghouse and a classified benchmarking process while explicitly prohibiting mandatory licensing or pre-clearance requirements. The framework is notable for what it withholds as much as what it establishes.
GitHub Copilot flipped to usage-based token billing on June 1, with developers immediately projecting cost spikes from $29 a month to over $750, and from $50 to over $3,000. The core complaint is that Microsoft spent months encouraging heavy usage under flat-rate pricing and is now repricing against its most engaged users. The backlash is shaping up as one of the sharpest developer community revolts of the year.
