Antigravity SDK v0.1.22: adds gemini-4-argon support and multi-agent workflows
The weekly Antigravity SDK update adds gemini-4-argon model targets (still defaulting to gemini-3.8-flash when unspecified), multi-agent workflows, in-memory skill definitions and output limits beyond 64k tokens.
Antigravity SDK October 12th 2026 weekly update for v0.1.22: • Gemini 4 Argon: The SDK now supports gemini-4-argon model targets. Still defaults to gemini-3.8-flash when unspecified. - Announcement: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/ • Dynamic Multi-Agent Workflows: Author and execute programmatic multi-agent pipelines with compile-time AST validation using @beta.workflows.define and agent.beta.run_workflow. • In-Memory Skill Definitions: Register programmatic agent skills directly in memory via InlineSkill and inline_skills without writing SKILL.md files to disk. • Extended Output Token Limits: Configure per-step output token budgets beyond 64k tokens via GeminiModelOptions.max_output_tokens for long-form reasoning and code generation. • Eager MCP Tool Registration & Payload Limits: Expose all tools from an MCP server eagerly via force_all_tools_eager and raise response byte thresholds via CapabilitiesConfig.mcp_max_output_bytes. https://github.com/google-antigravity/antigravity-sdk-python/discussions/255 🧵