Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.kymaapi.com/llms.txt

Use this file to discover all available pages before exploring further.

Google Antigravity is an agent-first IDE from Google DeepMind. It provides synchronized agentic control across your editor, terminal, and browser.

​
Official vs Community version

Official AntigravityOpen-Antigravity (community)
Custom API endpointsNot supportedSupported
Custom modelsGemini 3 Pro, Claude Sonnet 4.5, GPT-OSS onlyAny OpenAI-compatible model
Use with KymaNot possibleWorks out of the box
Downloadantigravity.googleGitHub
The official Google Antigravity does not support custom API endpoints or API keys. To use Kyma’s open source models with an Antigravity-like experience, use the community open-antigravity fork.

​
Setup with Open-Antigravity

1

Clone the repo

git clone https://github.com/ishandutta2007/open-antigravity.git
cd open-antigravity
2

Configure Kyma API

Create a .env file with your Kyma credentials:
PRIMARY_PROVIDER=openai
OPENAI_API_KEY=ky-your-api-key
OPENAI_BASE_URL=https://kymaapi.com/v1
3

Install and run

npm install
npm run dev
4

Start coding

Open the IDE and start using AI agents powered by Kyma’s open source models.
ModelAliasBest For
qwen-3.6-plusbestHighest quality overall
qwen-3-codercodeBest alias for code-focused tasks
kimi-k2.6agentBest tool calling, agentic workflows
gemini-2.5-flashlong-context1M context for large codebases
Use model aliases like best or code — they auto-update when better models become available.
curl "https://kymaapi.com/v1/models/recommend?agent=opencode"

​
Why use Kyma with Antigravity?

  • Active open models — access models from Meta, Google, Alibaba, DeepSeek, and more through a single endpoint
  • 4-layer fallback — if one provider is down, your request automatically routes to another
  • Prompt caching — up to 90% cost savings on repeated system prompts
  • Free to start — $0.50 signup credits, no credit card required

​
Prompt Caching

Prompt caching is automatic. Your system prompt and tool definitions are cached for 5 minutes, reducing costs by up to 90% on subsequent requests. No code changes needed. See Prompt Caching for details.