RT Max Weinbach: I had early access to this model and I'll post more Thursday, but I think the most impressive part is this model NEVER GIVES UP If yo...
Max Weinbach says he had early access to OpenAI's new model GPT-5.6 Sol, calling it his favorite model by far. He highlights that it never gives up and will keep reasoning until it's done. OpenAI announced that GPT-5.6 Sol, along with Terra and Luna, will launch publicly on Thursday, with preview access expanding globally now.
Background
- Paul Graham (YC co-founder) and Max Weinbach (leaker) are tweeting about a new OpenAI model called **GPT-5.6 Sol**, launching publicly on Thursday alongside two other variants, Terra and Luna.
- The post describes "Sol" as a reasoning model that perseveres until it solves a problem, never giving up — a notable improvement over earlier GPT models, which sometimes stop early or produce superficial answers.
- This appears to be a mid-cycle release (5.6, not 5.0 or 6.0), suggesting iterative upgrades rather than a full generational leap.
- The names Sol, Terra, Luna (Sun, Earth, Moon) hint at three specialized tiers: possibly one for heavy reasoning (Sol), one for balanced tasks (Terra), and one for lightweight/creative (Luna).
- The tweet matters because it signals OpenAI is racing to ship multiple increasingly capable models amid fierce competition from Anthropic (Claude), Google (Gemini), and open-source alternatives.
OpenAI introduced GPT‑Live, a new version of its AI model designed for real-time, interactive voice conversations. The model enables more natural, low-latency spoken interactions with users, marking a step toward more conversational AI capabilities.
Cognition has released SWE-1.7, a new version of its AI coding agent that they claim approaches near GPT-5.5 and Opus levels of intelligence, marking a significant advancement in autonomous software engineering capabilities.
Newer Claude models sometimes invent extra keys in tool call arguments, breaking validation in Pi's edit tool. The author suspects post-training for Claude Code's forgiving harness makes alternative schemas fail. This suggests closed RL training can degrade general tool-use reliability.
Meta has released Miles, a PyTorch-native open-source stack designed for large-scale reinforcement learning post-training of large language models. It provides a framework for training stability, reproducibility, and scalability, building on existing technologies like torchtitan and Fairscale to support distributed RL workflows.