Research lab
US-based AI research lab building open-source models, including its Hermes line, and tools for autonomous agents.
Nvidia launched RTX Spark laptops on Friday with up to 128GB of shared memory and a 74-billion-parameter model it says fits in 24GB. A local agent pilot is now possible on Windows, though every figure so far comes from Nvidia.
Reality
Nous Research raised $90 million at a $1.5 billion valuation to sell its open-source Hermes agent to corporate IT departments. The money pays for single sign-on and audit controls, so the price depends on companies paying for features around code anyone may copy.
Perspective Coverage
8 publishersReality
Lightpanda shipped 1.0 of its open-source browser on October 2nd, passing 1,739,845 web-platform subtests without rendering pages. One encoding suite makes up about two-thirds of that count, so teams should judge it by whether their jobs need to see the page.
Reality
Photon raised $4.5 million to help developers run AI agents inside iMessage and WhatsApp, citing over 40,000 developer sign-ups. Its open-source version still accounts for 98% of use, so the paying side of the bet is early.
Perspective Coverage
3 publishersReality
CARBONATO, a Docker botnet running since October 2024, hijacks hosts on open port 2375 and steals keys from 14 AI providers to fund its own LLM gateway. ThreatDown found the crew's own container registry exposed, handing defenders 4.3 GB of its toolchain.
Perspective Coverage
3 publishersReality
OpenClaw's Telegram enabled flag stopped inbound polling while cron-driven sends kept firing hourly through three stale bot accounts, one operator reports. The dashboard showed the channel off, and the sends ended only when the bot tokens left the secrets directory.
Reality
Manifold Security reported eight instances across seven command-line agents. The command runs as the user, outside the sandbox, with no approval prompt, and four were still firing when the researchers retested on September 1.
Reality
A health dashboard update from AWS reports customer data lost for good in the Middle East. The account carrying its wording to me is a Thai-language post that a model drafted and a human editor checked.
Reality
Ternary weights at 1.76 bits put a 27B model into a 5.9GB file and let a laptop decode it at 28.1 tokens a second. The retention figure comes from Prism ML's own benchmark suite, not the table on the model card.
Reality
OpenRouter has dropped the stealth listing. The model is Pareto 26.9, from unbiased.ai, at $2.50 per million input tokens before a 10 October launch. That is double what 26.8 charged.
Reality
A dev.to comparison scores MCP tool design at 13 website builders and content systems against eight criteria. The two designs it walks through in full both take deletion out of the flat tool list and put it behind its own gate.
Reality
A Thai-language walkthrough estimates 15 to 20 MCP calls to build a ceramics studio site on WordPress.com. Counting each documented operation, including one status update per item to publish, gets to 22.
Reality
One MIT-licensed proxy on localhost is enough to serve OpenAI's own desktop client from self-hosted models. The models that fail there fail on Codex's tool-call format. One of three tested got lost.
Reality
A dev.to post pulled every figure for 12 popular Claude Skills straight from the GitHub API on 19 September 2026, then set the result against what three skill catalogs advertise for the same repository.
Reality
Nous Research spent about $19,300 on 1,393 subagents to take 34.4 percent out of Hermes's non-test Python in 19 hours. Test line count moved 0.06 percent, so the guardrail was the suite the team already had.
Reality
Google's server for Google Analytics is a read-only Python package you run locally through pipx, and it has been on GitHub since mid-2025. Anthropic only handed the protocol to the Linux Foundation that December.
Reality
Mahidol's Biomedical and Data Lab publishes Thai Whisper fine-tunes that are free to use commercially, with 6.59 and 7.42 WER on Common Voice 13. Both were scored with the Deepcut tokenizer, and that shared segmentation is the reason the two numbers sit on one scale.
Reality
Nvidia's Nemotron scored 30 of 42 at IMO 2026 writing proofs in ordinary prose, and the September 9 paper publishes the data, the code and the acceptance rule that decided which candidate proofs survived.
Reality
PAIR cut a five-subagent inbox task from 18 minutes to 8 minutes 48 seconds across three machines, which is just over two times the speed for three times the hardware, and every box in NVIDIA's demo cluster was one NVIDIA sells.
Reality
A scan of 6,214 corporate domains found 120 sites whose AI-facing docs point at code packages and domains nobody owns, and registering a few of those names produced installs traced back to Claude, Codex and Hermes agents inside real networks.
Reality