AI coding assistants expand model choices and development workflows

The briefing · 2 Aug 2026 · 4 tracked changes

AI coding assistants expand model choices and development workflows

Recent updates to OpenAI's Codex and the coding platform Lovable show AI development tools introducing more model options, deeper system integrations, and expanded full-stack capabilities.

Codex updates models and integration options

OpenAI's coding tool Codex has added three new models to its system: GPT-5.6 Sol, GPT-5.6 Terra, and GPT-5.6 Luna. Alongside these model additions, the platform has expanded its integration features by adding Codex CLI, Codex IDE extension, and Codex cloud, raising its total capability count from five to seven.

Why it matters Developers are gaining more precise control over which underlying AI models they use and how they deploy them directly into their daily work environments.

codex

Built from 2 tracked changes

Lovable incorporates Gemini 3.6 Flash

The coding platform Lovable has added Gemini 3.6 Flash to its list of supported models. The tool already supports Claude Opus 4.5, GPT-5.2, and Gemini 3 Pro. This addition gives users another alternative for generating and editing code within the platform.

Why it matters Access to a wider variety of models within a single platform lets developers choose the best balance of speed, cost, and accuracy for their specific tasks.

lovable

Built from 1 tracked change

Lovable transitions to full-stack and GitHub syncing

Lovable has introduced three major new capabilities: an AI software engineer function, support for building full-stack web applications, and direct syncing with GitHub. These updates increase its total features from five to eight.

Why it matters AI code assistants are moving beyond simple code snippet generation to managing entire web applications and syncing automatically with standard developer hosting services.

lovable

Built from 1 tracked change

See the raw change log