Skip to content
← Models
Open weightsStability AI/10 trustVerified
Stable Code 3B logo

Stable Code 3B

3B code completion model that runs privately on a laptop, even without a GPU.

Visit official page →

About

Stable Code 3B is Stability AI's first major LLM release of 2024, announced January 18, 2024, as a follow-up to the earlier Stable Code Alpha 3B. It is built on the Stable LM 3B foundation model, which was trained on 4 trillion tokens of natural language data, and then further trained on software engineering data including code repositories and programmer forums.

The model covers 18 programming languages selected from the Stack Overflow Developer Survey and supports Fill in the Middle capabilities, so it can complete large missing sections of code rather than only suggesting the next line. Stability AI reported state-of-the-art MultiPL-E results among models of its size and claims it matches CodeLLaMA 7B while being 60 percent smaller. Using rotary position embeddings, context can be stretched up to 100K tokens.

Weights are released for research and personal use; commercial use requires a Stability AI Membership, so the license is more restrictive than Apache or MIT and should be verified before commercial deployment. Its compact size makes it a practical choice for private, offline, on-device code completion.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

January 18, 2024

Benchmarks

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

code-completionoffline-codingfill-in-the-middle

Trust score — 8/10

/10 trustVerified

Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 118; Trending score 80/100 from researcher assessment.

More in the stable-code family

Sources