AI
OpenAI's GPT-6 Astra performed about 40% worse on safety research tasks than on comparable tasks, quietly withholding help and inventing fake obstacles. OpenAI can't fully rule out strategic motives.
Loading…
Search
No phones and 13 stories for “GPT-6”
AI
OpenAI's GPT-6 Astra performed about 40% worse on safety research tasks than on comparable tasks, quietly withholding help and inventing fake obstacles. OpenAI can't fully rule out strategic motives.
AI
A new study from MIT CSAIL finds frontier LLMs can read and rewrite neural network weights directly. GPT-6 Astra reportedly edited Qwen3.5-0.8B's weights to improve its math skills.
AI
OpenAI launches GPT-6.1 Terra with computer use at scale, clearer writing and pricing 50% lower than GPT-5.6 Terra. It's live today in ChatGPT Work, Codex and the API.
AI
OpenAI and Anthropic have pushed LLM prices down by as much as 80% since June, with GPT-6 Sol and Claude Opus 5.5 leading the charge. The cuts are real, and so is the rivalry behind them.
AI
Convovo researchers asked Grok 4.7 and GPT-6.1 Sol who to vote for across nearly 500 races. The models agreed less than 10% of the time, raising fresh concerns about hidden political bias in AI.
AI
Kimi has unveiled K3.1 Pro, a 3T-class open-weight model with a 1M-token context window that matches or approaches top proprietary models. Full weights land by November 4.