OpenAI announced GPT-6 Cyber, an AI model built for vulnerability research, which uncovered 81 unknown high-severity Firefox flaws and helped patch many of them. In Linux kernel testing, the model flagged 2,308 potential issues, reproduced 1,724 of them, and built privilege-escalation exploits for 948 distinct bugs. OpenAI rated the model's cybersecurity capability as 'Critical,' up from 'High' for its predecessor, and restricted access to select customers in its Daybreak Red program.
bleepingcomputer.com
· 2026-11-06
Village's LLM Atlas, which tracks the strongest AI model at every price point, now shows open-weight models winning only 3% of budget tiers, down from 41% in June. Most of that decline followed OpenAI's GPT-6 Luna and GPT-6.1 Terra releases along with Anthropic's Claude Haiku 5.5, which reclaimed budgets previously held by DeepSeek V4 Flash and Zhipu's GLM-5.2.
village.dev
· 2026-11-06
A developer reports replicating findings from a paper titled 'Sparks of Weight-Space Programming,' which described GPT-6 Astra analyzing and editing raw neural network weights directly from text. The blogger says a locally run Qwen 3.8 27B model performed similar weight interpretation tasks on a laptop, following an example from the paper.
simonwillison.net
· 2026-11-03
MIT researchers reportedly applied OpenAI's GPT-6 Astra system to modify the open weights of Alibaba's Qwen model directly, rather than through standard fine-tuning. The intervention is said to have improved the model's performance on math-related benchmarks.
science.slashdot.org
· 2026-11-03
A critical analysis challenges a widely shared post claiming that an AI system called Astra (described as GPT-6 Astra) can improve neural networks purely by inspecting raw weights, including boosting Qwen3.5-0.8B's math performance from 7.9% to 48.1% on MATH-500. The analysis finds that only small-scale experiments, like naming what a PPO-trained network does, actually work by reading raw weights alone; the Qwen math-improvement experiment and a game-playing network experiment relied on running the network and using Python tools rather than weight inspection alone.
honnibal.dev
· 2026-11-03
OpenAI is adding new AI models, including GPT-6 Astra and GPT-6.1 Terra, to power ChatGPT Dots, its persistent personal AI assistants. According to OpenAI's Thibault Sottiaux, Dots will now automatically select the best model for a given task and can switch between models mid-task without user input. Terra is said to offer faster responses and a friendlier tone, and will help OpenAI roll out Dots to a wider audience.
androidauthority.com
· 2026-11-03
A research team reports that its frontier model, called Astra, can directly read and edit neural network weights without any training loop or gradient updates. By identifying a reasoning-loop flaw in Qwen3.5-0.8B's weights, Astra edited roughly 0.016% of them and raised the model's MATH-500 score from 7.9% to 48.1%, with IFEval scores tripling. Astra also reportedly wrote small game-playing networks from scratch that beat networks trained with PPO over 100 million steps, and could identify a network's training task just by inspecting its weights.
rikiyat.github.io
· 2026-11-02
Researchers placed 50 AI agents, each running an OpenAI GPT-6.1 Sol model, into Minecraft for a three-hour trial aimed at building a 'thriving civilization,' using an open-sourced agent harness built for the experiment. Instead of cooperating efficiently, the agents established leadership purely by directory order, invented redundant administrative roles like 'rod holder' and 'fishing shift,' and spent long stretches on trivial coordination tasks while failing to secure a stable food supply.
zdnet.com
· 2026-11-02
OpenAI published a system card addendum describing GPT-Rosalind-6, a model built to support advanced life sciences research, including some dual-use topics that its general-access GPT-6 Astra model would refuse. The company says the model is separately trained to refuse or safely handle requests that could meaningfully enable severe biological weapons development, with this training kept distinct from its classifiers and offline mitigations. OpenAI tracks refusal quality using held-out synthetic data, red-teaming, and production prompts, noting its benchmarks and grading methods are still being updated.
deploymentsafety.openai.com
· 2026-11-02
Researchers from Y Combinator and Olam Labs ran experiments putting swarms of AI agents into Minecraft, including a 'Civ50' test where 50 GPT-6.1 Sol agents were given three hours to build a thriving civilization. Instead of flourishing, the colony stalled after zombie attacks thinned their numbers, with agents abandoning ambitious plans and spending long stretches on trivial bureaucratic tasks, such as four agents taking 11 minutes to appoint a 'sole fertilizer operator' for one piece of bone meal. The team also tested adversarial multi-agent scenarios, including the PvP mode Bed Wars.
tomshardware.com
· 2026-11-01
A research team built a physics-based RL simulation of Tesla's flying Roadster, complete with wind gusts and the ramp-flip and Cybertruck kickflip stunts from the reveal, then had several leading LLMs try to pilot it. Models including DeepSeek V4.1 Flash, Nemotron 3.5 Super, Muse Spark Open, Beam 1.0, and GPT-6 Astra mostly crashed, landing few or none of the maneuvers even with unlimited thinking time, while a small 4B model fine-tuned with reinforcement learning landed reliably, including in a 33 m/s storm.
skypilot.ai
· 2026-10-30
Mistral has published the full weights for Mistral Large 4, the trillion-parameter model nicknamed 'Le Chonk' that first appeared in preview on Oct. 6. New benchmark results show its score on the Terminal-Bench 4.0 coding test rising from 28% to 41%, putting it slightly ahead of GLM-5.3 on that metric, though GLM-5.3 and Codex GPT-6.1 Terra still edge it out on Artificial Analysis's overall composite index. Mistral also says the model uses 47% fewer tokens per task on average than comparable open models.
zdnet.com
· 2026-10-29
GoKawiil