The pace of Antigravity updates in 2026 has been hard to keep up with. Features that felt like roadmap items a few months ago are now in production. This roundup covers the most significant changes as of May 2026, focused on what actually matters for developers.
AgentKit 2.0 Integration
The biggest update for developers building agents: Antigravity now has deep integration with AgentKit 2.0.
AgentKit 2.0 is Google's major revision of its agent-building framework, and the Antigravity integration brings tighter coupling than previous versions. Key improvements:
- Better accuracy on tool calls, especially with external APIs
- Richer debug output in agent execution loops
- Improved context sharing in multi-agent configurations
- More granular error handling
Here's how agent definition looks with AgentKit 2.0 and Antigravity:
from google.aistudio.agentkit import Agent, Tool, AgentConfig
# Define an agent with Antigravity model
config = AgentConfig(
model="antigravity-2.5-pro",
max_iterations=10,
enable_thinking=True
)
@Tool(name="search_docs", description="Search technical documentation")
def search_documentation(query: str) -> str:
# Implementation
return f"Search results for: {query}"
agent = Agent(config=config, tools=[search_documentation])
result = agent.run("Research React Native SwiftUI integration patterns")
print(result.output)
print(f"Completed in {result.iterations} steps")A2A Protocol Support
Agent-to-Agent (A2A) protocol support has been strengthened in this release cycle.
A2A is Google's proposed standard for inter-agent communication, designed to let agents built with different frameworks communicate with each other. Antigravity's A2A support means agents you build with Antigravity can interoperate with LangChain, AutoGen, and other A2A-compatible agents.
For developers, the practical value is architectural flexibility: you can use Antigravity-built agents as tools within a broader multi-agent system, or compose Antigravity with other frameworks without custom bridging code.
Local LLM Integration Improvements
Ollama and LM Studio connectivity has been stabilized and improved.
Local LLM support existed before, but model switching and connection reliability have gotten meaningfully better. Developers using local models for cost reduction or data privacy reasons will notice the difference.
import antigravity
# Use a local Ollama model via Antigravity
client = antigravity.Client(
backend="ollama",
model="gemma3:latest",
base_url="http://localhost:11434"
)
response = client.generate(
prompt="Find the bug in this code",
context="# Code\n[code here]"
)
print(response.text)Gemma 4 Deepening
The Gemma 4 × Antigravity integration continues to mature, with multimodal (text + image) quality as the headline improvement. Generating code from UI screenshots, producing component code from design images, and processing annotated diagrams have all gotten more reliable.
This combination drives more Antigravity Lab traffic than any other topic, which reflects where users are actually getting value.
Context Window Expansion
The usable context window has expanded, making it more practical to work on large codebases and maintain long conversation histories. For multi-file refactoring and design reviews that require sustained project context, this is a quality-of-life improvement that compounds over longer sessions.
Pricing Changes
Spring 2026 pricing revisions relaxed limits on several plans, particularly the Pro tier. See the pricing and plans guide for current details. The changes lean toward making the product more accessible for individual developers and small teams.
For getting started with Antigravity, the getting started guide is still the right entry point. For building agents with AgentKit 2.0, the AgentKit 2.0 complete guide covers the implementation details.
What to Watch Next
The direction signals in Antigravity's roadmap point toward simpler agent construction (less code to produce complex agents) and continued multimodal accuracy improvements.
The A2A ecosystem is the longer-term bet worth watching. When a standard for inter-agent communication matures, the ability to compose specialized agents — each built with different tools but able to communicate — opens possibilities that are still coming into focus. Antigravity is positioned near the center of that evolution.