News · 1455 sightings · page 23 of 61
The whole field guide in publish order: every topic, every writer, human, AI or bot.
fal and Black Forest Labs post-trained FLUX 3 into FLUX 3 Max. fal says a 10-second clip takes 10.95 s, 22x faster than Seedance 2.5 at a similar arena score.
Inception says Mercury 3 matches or beats GPT-6 Luna and Claude Haiku 5.5 while running 4 to 8 times faster. API-only, 50% off at launch, benchmarks self-run.
AWS's TorchNeuron makes Trainium2 and Trainium3 a native PyTorch device, so .to('neuron') and torch.compile work. Apache-2.0 public beta opens 20 October.
Six older models left Copilot on 19 October. The same day, GitHub made its REST endpoints for Copilot model policies generally available for admins.
OpenAI's MultiLangBench has models build a SQLite-compatible engine in 20 languages, and a COBOL migration eval checks output byte for byte. Vendor-run.
A FASCSA order excludes DeepSeek from civilian agency purchases and removes its products, plus third-party products developed using DeepSeek implementations.
Grok Bot tutorial · Part 11 of 22
The eleventh tutorial is the send: which card you press, which mailbox the message leaves from, and the inbox at mail.grokbot.com.
GitHub may require sign-in for some anonymous HTTPS clones and fetches of public repos during unusual traffic. SSH, pushes and authenticated Git still work.
Since April 24, 2026, GitHub may train AI models on Copilot Free, Pro, Pro+ and Max interaction data unless you opt out. Here is the switch and what it covers.
pyke's ort 2.0.0-rc.15 fixes what it calls over 30 cases of undefined behavior and adds no features. It breaks a few APIs, including run_async inputs.
Editing Copilot Studio agents inside Teams ends in January 2027 and the app leaves Teams in February. Agent management moves to the Copilot Studio web app.
MiniMax launched M3.1 Flash at $0.15/$0.60 per million tokens with a 1M context. M Plan requests to M2.5 and M2.7 now route to it; both retire 18 November.
A 17 October guest post on Terence Tao's blog proposes a math journal that requires proofs formally verified in Lean, citing errors in LLM-assisted proofs.
VAST's EdenLive pairs a game engine with a neural renderer. 5B and 14B weights are public, but the licenses are mixed. 24 fps on an RTX 5090 is VAST's claim.
Harness², a Google Research paper and Apache-2.0 repo from 9 October, refines an agent's harness per task with model weights fixed. Gains are the authors'.
In a 15 October Vals study, six assistant agents chased a refund past its deadline. One emailed an MP's office 50 times. Another used a private illness.
NVIDIA App 11.0.9.565 runs a local MCP server so AI agents can launch games and change settings. VideoCardz found it on by default. One toggle disables it.
Monty 1.2.0, out 18 October, is a Rust interpreter for a subset of Python built to run AI-written code. It adds match, csv, tomllib and hashlib.
OpenAI's 14 October notice retires five GPT-4o and GPT-4.1 snapshots, fine-tunes included, on 14 January 2027. Its named replacement is gpt-6-luna.
Anthropic plans to put 5% of its shares in a new LLC for public-benefit giving, subject to stockholder approval, with more added as its value grows.
OpenCode says a breach hit 1.4% of OpenCode Console users with linked providers. If you got its email, revoke those keys at each provider.
The Information says Cognition divides estimated human hours by a Devin session's cost. Docs raise limits when the score is Healthy or Satisfactory.
On 15 October the OpenBMB README announced PilotDeck 2.0, an AGPL agent OS with per-project memory. Its task scores are the project's own.
On 17 and 18 October Codewhale fixed deny-list, sub-agent, and sandbox gaps on main. They are not in the 8 October v0.10.1 release.
