アイウ01エオカ
キクケコサシスセソ01
タチツテトナニ
ヌネノハヒフヘ01ホ
マミムメモ
ヤユヨラリルレロ01ワ
ヲン01アイ
ウエオカキクケコ
サシスセソタ01チ
ツテトナニヌネ
ノハヒフ01ヘホマ
ミムメモヤユ
ヨラリルレ01ロワヲ
ンアイウエ
オカキクケコサ01シ
スセソタチツテト
ナニヌネノ01ハ
ヒフヘホマミムメ
モヤ01ユヨラ
リルレロワヲン01ア
イウエオカキ
クケコサ01シスセ
ソタチツテト
ナニヌ01ネノハヒ
フヘホマミムメ01
Models built for agents

Models built for coding agents

Source control with out-of-the box codebase retrieval, fast utility SLMs, and task-specific agents you can run on any repo.

Trusted by the best leading brands

ACME
VORTEX
NEXUS
PRISM
HELIX
FLUX
terminal
$relace init
Initializing workspace...
$relace push
Indexing repository ████████████ 100%
Generating embeddings... done
$
Features

Everything you need for autonomous codegen

Purpose-built infrastructure that lets your agents write, retrieve, and merge code without bottlenecks.

Repos

Source control designed for agents with light weight push/pull operations and no rate limits.

Code Retrieval

Best in class semantic search that scales to large codebases out-of-the-box.

Fast Apply

Universal code merging model that applies file edits at 10,000 tok/s.

Online
Latency
12ms
100%75%50%25%0%RETRIEVAL ACCURACY OVER TIME99.7%
Throughput
10,000 tok/s
Models

State of the art SLMs as tools for coding agents

Small, fast models trained in-house to outperform frontier LLMs on utility tasks. Equip your agent with tools to apply file edits at 10,000 tok/s, and search an entire codebase in less than 2s.

Fast Apply
10k tok/s
Search
<2s
Accuracy
99.7%
Uptime
99.99%
Infra

Source control designed for the models using it

Lightweight push/pull operation from sandboxes, fast branching for spawning subagents, automatic indexing for two-stage retrieval, and rate limits designed for high throughput.

Sandbox isolationFast branchingAuto-indexingNo rate limits
Branch Activity
Live
8f4a2b1HEAD
feat: add semantic search
Agent-01
2m ago
3c9d7e4branch
fix: merge conflict resolution
Agent-02
15m ago
1a5b8c2
refactor: optimize indexing
User
1h ago
d7e9f01
ci: update pipeline config
CI/CD
2h ago
Features

Building blocks for reliability and scale

Source control with out-of-the box codebase retrieval, fast utility SLMs, and task-specific agents you can run on any repo.

01

Specialized models

Custom-trained SLMs for code retrieval, merging, and generation.

02

Fast retrieval

Semantic search across entire codebases in under 2 seconds.

03

Smart merging

Context-aware code merging at 10,000 tokens per second.

04

Low latency

Optimized inference stack with sub-100ms response times.

05

Simple integration

Drop-in API with SDKs for every major language and framework.

06

Built for reliability

SOC 2 compliant infrastructure with 99.99% uptime SLA.

Testimonials

Trusted by trailblazers

Relace has been a critical tool for us to create custom, design-focused AI models and allowed us to fine tune on our own data and continuously create better and better models.

TN
Teddy Ni
Co-Founder at Magic Patterns

Relace’s fast rewriting model has been a big boon to our product. It’s made edits more reliable with hardly any visible downsides. A whole class of bugs is gone for us now.

JG
James Grugett
Co-Founder & CEO at Codebuff
FAQs

Frequently asked questions

Relace is purpose-built for coding workflows. Instead of relying on general-purpose LLMs, our in-house models specialize in retrieval, merging, and code generation — making them faster, more reliable, and easier to integrate into engineering pipelines.

Get started in minutes

Try out our playground, and start with our free tier to test Relace models in your application.