Home
Fast-tracking AI industry hot events with concise ~1000-word analysis.
On 2026-07-31 ByteDance officially released Seedance 2.5 on Doubao Pro and Dreamina. Three core upgrades: up to 30-second single clips (up from 15s), up to 50 all-modal material fusion, and 30s native single-take output. Narrative capability leaps from clip stitching to single-take complete creation. Seedance 2.0 was previously world #1.
Two US House committee chairs - Rep. John Moolenaar (Select Committee on the CCP) and Rep. Andrew Garbarino (Homeland Security Committee) - sent a joint letter to DoorDash CEO Tony Xu investigating the company's use of Chinese AI model Kimi K2.6. DoorDash's internal AI lab had adopted the model after finding it outperformed US counterparts on certain tasks, but the decision collided with national security and data security concerns. The case exposes the fundamental tension between performance-driven engineering selection and regulatory risk logic - when the best model comes from a strategic competitor, these two frameworks cannot auto-align.
SpaceXAI released Grok 4.5 on July 8, 2026 - a 1.5T V9 model co-trained with Cursor, its first built specifically for coding and agents, priced $2/$6 with a 500K context and pitched internally as "comparable to Opus 4.7 but much faster"; no public benchmarks exist, so it positions as a cheap Opus-class substitute native to Cursor, differentiated from Opus 5 / GPT-5.6 Sol on ecosystem rather than leaderboard rank.
On July 9, 2026 Zuckerberg returned to X to launch Muse Spark 1.1, a 1M-context agentic model at $1.25/$4.25 with Meta's first paid API. It leads agentic tool-use benchmarks (MCP Atlas 88.1, JobBench 54.7) but its independent Intelligence Index is just 51, with coding and long-horizon GDPval-AA v2 trailing; two weeks later Opus 5 and GPT-5.6 pushed it down the field. Its real edge is token efficiency at a rock-bottom price (~$0.26/task).
On 2026-07-31 DeepSeek launched the official (stable) V4-Flash API to public beta; the model name stays deepseek-v4-flash, with the same architecture as Preview, only re-post-trained. Agent capability is greatly enhanced, with official benchmarks far exceeding V4-Pro-Preview (Terminal Bench 2.1 82.7, Cybergym 76.7, DeepSWE 54.4, etc.). It natively supports the Responses API and is adapted for Codex; the V4-Pro official version is coming next.
Anthropic released Claude Opus 5 on July 24, 2026, topping both the Artificial Analysis Intelligence Index (61) and Agentic Index (55.3), priced at $5/$25 unchanged from Opus 4.8. Its real edge is agentic capability (leading AA-Briefcase/GDPval), but it runs slow, burns tokens, and its hallucination rate rose to 50%; it is built for complex long tasks, not everyday Q&A.
V8.1 launched April 2026 and served as Midjourney's default model through summer, with video animation tools arriving in the July update - it turns high-quality stills into 5-second motion clips, filling the 'make it move' gap rather than replacing Sora as the best video tool.
OpenAI's own report shows Codex went from novelty to mainstay in six months-Research usage up 56x, 70.2% of sampled users ran ≥1-hour tasks, 25.6% over 8 hours. Non-developers are now driving enterprise agent growth, and long tasks fit agents only when boundaries are clear.
Mistral AI ships its first embodied model Robostral Navigate, an 8B model that navigates with a single RGB camera and hits 76.6% on the R2R-CE benchmark, beating multi-sensor stacks.
Meituan open-sources LongCat-2.0, a 1.6T sparse-MoE coding model trained entirely on Chinese ASIC with zero Nvidia GPUs, that secretly topped OpenRouter under the codename Owl Alpha for about two months. It beats Qwen3.6-27B across coding benchmarks, with three takeaways for everyday developers.