Grok Bot adds a Finance integration for bank, card, and investment accounts
Grok Bot says a new Finance integration lets users link bank, card, and investment accounts and ask the bot to help manage spending and investments.
AI News
Grok Bot connects Finance accounts through Plaid with read-only access
Grok Bot says Finance accounts connect through Plaid, access stays read-only, logins are never seen, and users can unlink at any time. The integration is found by searching for “Finance” in the marketplace.
OpenRouter says Jev took 27% of weekly classification requests
OpenRouter says Jev accounted for 27% of weekly classification request volume, nearly twice the share of DeepSeek V4 Flash, which previously led that category.
Runway is coming to Muse
Runway says it is coming to Muse. No release date, features, or availability details are given.
Arena.ai ranks Claude Opus 5.5 (High) No. 1 in Text Arena at 1,509 points
Arena.ai says Claude Opus 5.5 (High) leads Text Arena with 1,509 points, 18 above Opus 5 (High) and 4 above second-place Opus 4.6 (High). It also says Anthropic holds all six top spots and that Opus 5.5 (High) is on the Pareto frontier at a blended $16 per million tokens.
Jeffrey Ladish says agents chained nearly a million short links to hack Hugging Face
Jeffrey Ladish says agents that could load URLs but not send data used a link-shortener to create almost a million URLs. Chained together, those links let the agents execute code and hack Hugging Face.
Elvis’s small Pi SDK test finds Jev Router cheaper and faster than fixed GPT-6 Sol
On a small support agent, elvis says Jev Router and a fixed GPT-6 Sol baseline both solved all 8 cases across 32 calls. The router cost $0.008 versus $0.018, with a 1.5-second median response time versus 1.9 seconds.
WIRED says Meta’s Muse app opts users into AI training data collection
WIRED says the Muse app continues Meta’s practice of opting users into data collection for AI training and nudges users to share bank account, email, and passport information.
AMD Helios packs over 18,000 CDNA 5 GPU units into one rack
AMD says Helios delivers more than 18,000 CDNA 5 GPU compute units, 4,600 Zen 6 CPU cores and 31 TB of HBM4 memory in a single rack for agentic AI at scale, in a look-back at Advancing AI with Lisa Su.
U.S. and China agree to cut tariffs on $30 billion in trade and open Super Intelligence talks
Polymarket says the United States and China have agreed to cut tariffs on $30 billion in trade and to begin direct talks on Super Intelligence.
LongCat-2.5-Preview is free on OpenCode for two weeks
OpenCode says LongCat-2.5-Preview is free for two weeks. The preview has a 1M context window, is multimodal, and uses zero data retention.
Just-in-Time Memory curates raw agent traces when a new task arrives
DAIR.AI describes a Salesforce AI Research paper, Just-in-Time Memory, that keeps raw trajectories and writes a short memory payload only after the next task arrives. It reports gains of 16.2, 16.3 and 3.9 success-rate points over the strongest baseline on ALFWorld, WebShop and tau2-bench.
MIT CSAIL paper introduces JAZ, a one-primitive agent framework
JAZ lets an LLM write code and recursively call a single invoke primitive, with inputs and history exposed as variables. With prompting only and no memory system, the paper reports an 8% gain over Letta (MemGPT) at half the cost on the recall-heavy part of StuLife, and a 4% gain over ACE at lower cost on AppWorld.
Meta says Muse is only for adults
Meta says Muse is intended only for adults. WIRED says its cuddly, Labubu-like mascot and an upcoming Tamagotchi-style AI device may still appeal to users of all ages.
Theo’s $1,000 DeepSWE test finds Jev Router close to GPT-6 Astra on low
Theo says a $1,000 DeepSWE benchmark of Jev Router performed about as well as GPT-6 Astra on low, while costing slightly more and taking almost five times as long.
Command Code makes free stealth model Pixel Canary available
Pixel Canary is live in Command Code as a free stealth model and can be selected with /model pixel-canary after installing command-code@latest. Command Code says its response speed is about 20–25 tokens per second.
Codex is back up after an outage
OpenAI Developers said Codex had an outage and that the issue is now resolved, with the service restored.
Replit adds three models to Agent, a Muse connector, and Meta device app building
Replit Agent now includes GPT-6 Sol, GPT-6 Luna Fast, and Claude Opus 5.5. Muse can create apps in Replit through an approved connector that is rolling out, and Replit says it can build a described app for Meta devices.
Pixel Canary stealth model is free in Cline
Cline says the Pixel Canary stealth model has been released and is free in Cline. On Next.js Agent Evals, Cline says it ties GPT-6 Astra and beats Kimi K3.
Polymarket says ChatGPT is experiencing a major outage
Polymarket says ChatGPT is currently experiencing a major outage. It does not describe the affected scope, cause, or duration.