Kimchi
Run AI, not infrastructure.
Subscription555 impressionslaunches Jul 4
Kimchi is a managed AI inference platform that lets teams deploy, govern and scale LLM-powered coding agents via a single OpenAI-compatible API.
- for
- Engineering teams that need AI-assisted development with governance and cost controls.
- pricing
- freemium
- license
- Apache-2.0
Key features
- Spend limits — Hard caps can be set per user, team, API key or model and requests stop when the limit is reached.
- Multi-model routing — Automatically selects the best open-source or open-weight model for each coding step.
- Serverless inference API — One endpoint for many models with ~80 ms latency and no infrastructure to manage.
- Kimchi Studio — Visual workspace to watch, guide and resume AI agents.
- Teleport handoff — Transfer a terminal session to the cloud, continue processing, and retrieve results later.
- Workspaces — Persistent cloud machines for agents, built from admin-controlled templates.
- Team budget monitoring & RBAC — Real-time cost attribution per developer, team, model or tag with role-based access.
- Flexible deployment — Run serverless or self-host inside a VPC on AWS, GCP, Azure or on-premises.
Use cases
- Enforce per-developer spend caps while running AI code generation.
- Automated pull-request code reviews with inline comments and severity scores.
- Run AI coding agents serverlessly without managing GPUs.
- Deploy the platform inside a VPC for full data sovereignty.
- Hand off long-running terminal sessions to the cloud and retrieve results.
Kimchi pricing
- Coder$20/mo/monthIncludes $25 monthly credits · Priority request speed · 1M context window · Max 3 seats
- Teams$35/seat /mo/monthIncludes $45 monthly credits · Priority request speed · Kimchi Studio · Team budget monitoring and RBAC
- Code Review add-on$15/user /mo/monthAI PR review for GitHub/GitLab · Model choice per repo · Findings count against budget
Kimchi FAQ
Which models can I run with Kimchi?+
Kimchi’s multi-model harness stacks the strongest open-source and open-weight models.
Can I switch plans anytime?+
Yes, you can upgrade or downgrade at any time and changes apply immediately.
What payment methods do you accept?+
Major credit cards and cloud provider marketplace billing; enterprise customers can request invoicing.
Do you have any special offers for startups?+
Eligible companies can join the Startup Program; contact Kimchi for details.
How does the Open Source plan work?+
You download the CLI, run it on your own infrastructure, and bring your own inference or use Kimchi’s managed API.
Summarized by DevHunt from kimchi.dev · Oct 28, 2026. Details may change; check the official site.
