ChatGPT first released
OpenAI launches ChatGPT as a research preview. The moment LLMs crossed from research demo to mainstream utility, becoming the fastest-adopted consumer product in history.
New primitives are coming
Projects
Timeline
OpenAI launches ChatGPT as a research preview. The moment LLMs crossed from research demo to mainstream utility, becoming the fastest-adopted consumer product in history.
Anthropic previews Claude Code for agentic coding workflows. First credible move from "AI that suggests code" to "AI that does the work end-to-end."
OpenClaw first appears publicly as an open-source local agent project. Showed the value at the intersection of self-improvement, broad permissions, and easy access in a single agent.
Anthropic ships Claude Opus 4.6. Introduced "agent teams" and a 1M-token context, pushing coding agents from single runs to sustained, multi-agent work.
OpenAI introduces GPT-5.3-Codex. Set a new SWE-Bench Pro high and was the first OpenAI release flagged as "High" capability in cybersecurity.
Alibaba officially releases Qwen3.5. Made frontier-tier intelligence something you can download, not just rent — reshaping the open-vs-closed balance.
Google Research introduces TurboQuant. Compressed KV caches 6× with near-zero accuracy loss and no retraining, easing the memory bottleneck behind long-context inference.
Google launches Gemma 4. A 31B open model that matches 400B-class rivals, with multimodal reasoning running fully offline on phones and Jetson-class hardware.