# StableLM (model family)
StableLM is a family of open-weight large language models released by [Stability AI](https://www.edgepedia.org/stability-ai), beginning in April 2023. It progressed from 3B and 7B "Alpha" base models through the September 2023 StableLM-3B-4E1T to the Stable LM 2 series (1.6B, and reportedly 12B) in 2024, after which the family went dormant.
| Key fact | Detail |
|---|---|
| First release | StableLM Alpha, 3B and 7B parameters, April 2023, with 15B–65B models announced to follow [1](https://stability.ai/news-updates/stability-ai-launches-the-first-of-its-stablelm-suite-of-language-models) |
| Original training data | An experimental dataset built on The Pile but three times larger, with 1.5 trillion content tokens [1](https://stability.ai/news-updates/stability-ai-launches-the-first-of-its-stablelm-suite-of-language-models) |
| StableLM-3B-4E1T | 3B-parameter decoder-only transformer trained on 1 trillion tokens for 4 epochs, September 2023 [2](https://github.com/stability-AI/stableLM/) |
| Stable LM 2 1.6B | ~2 trillion tokens over two epochs, multilingual (7 languages), announced January 2024 [3](https://stability.ai/news-updates/introducing-stable-lm-2) |
| Vendor benchmark claim | State-of-the-art open model under 2B parameters at the time of the February 2024 technical report [4](https://arxiv.org/html/2402.17834v1) |
| Licensing | CC BY-SA-4.0 on 2023 base weights; Stability AI Community License on Stable LM 2, with commercial use requiring a separate Stability AI license [1](https://stability.ai/news-updates/stability-ai-launches-the-first-of-its-stablelm-suite-of-language-models) • [5](https://huggingface.co/stabilityai/stablelm-2-1%5F6b) |
| Status in 2026 | Dormant; no new models since April 2024, and third-party trackers describe the family as deprecated [6](https://recatools.com/ai-directory/stablelm/) |
## Releases and versions
The first public release, in April 2023, was the StableLM Alpha in 3 billion and 7 billion parameter sizes, described by Stability AI as the first of a "Stable LM suite" with 15 billion to 65 billion parameter models to follow [1](https://stability.ai/news-updates/stability-ai-launches-the-first-of-its-stablelm-suite-of-language-models). [Ars Technica](https://www.edgepedia.org/ars-technica)'s contemporaneous report confirmed the alpha's availability on GitHub in the 3B and 7B sizes and the promise of 15B and 65B models [7](https://arstechnica.com/information-technology/2023/04/stable-diffusion-for-language-stability-launches-open-source-ai-chatbot/).
In September 2023 Stability released StableLM-3B-4E1T, a 3-billion-parameter base model pre-trained on 1 trillion tokens for four epochs, a multi-epoch regime the model card says was designed to study the impact of repeated tokens on downstream performance, following Muennighoff et al. (2023) on data-constrained scaling [2](https://github.com/stability-AI/stableLM/). The [Hugging Face Transformers](https://www.edgepedia.org/hugging-face-transformers) documentation describes it as a decoder-only base model trained on "diverse English and code datasets" [8](https://huggingface.co/docs/transformers/model_doc/stablelm).
Stable LM 2 1.6B was announced in January 2024 as a multilingual small language model trained on approximately 2 trillion tokens for two epochs, covering English, Spanish, German, Italian, French, Portuguese, and Dutch [3](https://stability.ai/news-updates/introducing-stable-lm-2). The accompanying technical report appeared in February 2024 [4](https://arxiv.org/html/2402.17834v1). A 12B model released in April 2024 is reported by a third-party directory as the family's last model; the sources in this record do not include a primary announcement for that release, so its details rest on weaker sourcing [6](https://recatools.com/ai-directory/stablelm/).
## Architecture and training as published
StableLM's published models follow the LLaMA-style decoder-only transformer design. The StableLM-3B-4E1T card specifies 2,795,443,200 parameters, hidden size 2560, 32 layers, 32 heads, and a 4096 sequence length, with rotary position embeddings applied to the first 25% of head dimensions and LayerNorm rather than RMSNorm [2](https://github.com/stability-AI/stableLM/). The [Transformers](https://www.edgepedia.org/transformers) documentation adds SwiGLU activation and partial Rotary Position Embeddings [8](https://huggingface.co/docs/transformers/model_doc/stablelm). Stable LM 2 1.6B has 1,644,417,024 parameters, hidden size 2048, 24 layers, 32 heads, and a 4096 sequence length [4](https://arxiv.org/html/2402.17834v1), and was trained in bfloat16 with AdamW using the Arcade100k tokenizer, extended from tiktoken's cl100k_base, with a vocabulary of 100,352 [5](https://huggingface.co/stabilityai/stablelm-2-1%5F6b).
Training data changed markedly between generations. The April 2023 Alpha was trained on an experimental dataset built on [The Pile](https://www.edgepedia.org/the-pile) but three times larger, with 1.5 trillion content tokens [1](https://stability.ai/news-updates/stability-ai-launches-the-first-of-its-stablelm-suite-of-language-models). The September 2023 3B-4E1T model instead used a filtered mixture of open datasets from the [Hugging Face Hub](https://www.edgepedia.org/hugging-face-hub): the Falcon RefinedWeb extract, RedPajama-Data, and The Pile (both without the Books3 subset), and [StarCoder](https://www.edgepedia.org/starcoder) [2](https://github.com/stability-AI/stableLM/). Stable LM 2 1.6B kept that mixture and added multilingual data from CulturaX and its OSCAR corpora [5](https://huggingface.co/stabilityai/stablelm-2-1%5F6b).
## Benchmarks: vendor claims versus independent results
All quantitative benchmark claims in the public record are vendor-reported. Stability's launch post claimed the 3B–7B Alpha models showed "surprisingly high performance" in conversational and coding tasks despite their small size relative to GPT-3's 175B parameters [1](https://stability.ai/news-updates/stability-ai-launches-the-first-of-its-stablelm-suite-of-language-models). For the 3B-4E1T model, the company claimed state-of-the-art 3B-scale open-source performance as of September 2023, outperforming its own 7B StableLM-Base-Alpha-v2 [2](https://github.com/stability-AI/stableLM/).
For Stable LM 2 1.6B, the technical report's authors stated it was "the state-of-the-art open model under 2B parameters by a significant margin" at the time of publishing [4](https://arxiv.org/html/2402.17834v1). The launch post claimed it outperformed models under 2B parameters such as Phi-1.5, Phi-2, TinyLlama 1.1B, and [Falcon](https://www.edgepedia.org/falcon-model-family) 1B on most [Open LLM Leaderboard](https://www.edgepedia.org/open-llm-leaderboard) tasks [3](https://stability.ai/news-updates/introducing-stable-lm-2). On the vendor's MT-Bench table, stablelm-2-zephyr-1.6b scored 5.42, ahead of gemma-2b-it (5.19), phi-2 (4.29) and TinyLlama-1.1B-Chat (3.46), but well behind Mistral-7B-Instruct-v0.2 (7.61) and Llama2-Chat-70B (6.86) [4](https://arxiv.org/html/2402.17834v1). The report also stated that the instruction-tuned stablelm-2-1-6b-dpo improved on Microsoft's Phi-1.5 by two average points on few-shot accuracy while lagging behind the larger Phi-2.0 [4](https://arxiv.org/html/2402.17834v1).
Independent verification is thin in the available record. No third-party leaderboard or benchmark source for StableLM was retrieved, so the per-token efficiency and state-of-the-art claims were never independently confirmed here. A retrospective directory account says the April 2023 Alpha launch "landed with a thud", with early community testers finding the models rambling and weak against contemporaries [6](https://recatools.com/ai-directory/stablelm/), and by 2026 the same source places Qwen's small models, Gemma, Phi, and [Llama](https://www.edgepedia.org/llama-model-family) 3.2 ahead of StableLM with active support [6](https://recatools.com/ai-directory/stablelm/).
## Licensing and availability
The 2023 base models were released under CC BY-SA-4.0, permitting commercial and research use, while the fine-tuned instruction models, built on Alpaca, GPT4All, Dolly, ShareGPT, and HH datasets, were research-only under CC BY-NC-SA 4.0 [1](https://stability.ai/news-updates/stability-ai-launches-the-first-of-its-stablelm-suite-of-language-models). This split is the closest documented answer to why some early weights were restricted: the base weights were open, the derivative fine-tunes were not.
Stable LM 2 changed regime. The [Hugging Face](https://www.edgepedia.org/hugging-face) model card lists the Stability AI Community License and directs commercial users to a separate Stability AI commercial license [5](https://huggingface.co/stabilityai/stablelm-2-1%5F6b). The January 2024 launch post said both commercial and non-commercial use required a Stability AI Membership [3](https://stability.ai/news-updates/introducing-stable-lm-2); a later third-party description has the Community License free for individuals and companies under US$1 million revenue, with paid enterprise terms above that [6](https://recatools.com/ai-directory/stablelm/). These two accounts of the commercial terms differ and the record does not resolve them. The practical effect either way is that Stable LM 2 was not under a permissive open-source license such as Apache 2.0, a gap between the open-weight promise and the actual usage terms.
## Controversies and the decline of the LLM effort
The Alpha launch's weak reception set the tone for the family's standing [6](https://recatools.com/ai-directory/stablelm/). CEO [Emad Mostaque](https://www.edgepedia.org/emad-mostaque) resigned in March 2024, and the rescue investment that followed refocused Stability AI on image, video, and audio generation; the language models were never picked back up [6](https://recatools.com/ai-directory/stablelm/).
## References
1. [Stability AI Launches the First of its Stable LM Suite of Language Models](https://stability.ai/news-updates/stability-ai-launches-the-first-of-its-stablelm-suite-of-language-models)
2. [Stability-AI/StableLM (GitHub)](https://github.com/stability-AI/stableLM/)
3. [Introducing Stable LM 2 1.6B — Stability AI](https://stability.ai/news-updates/introducing-stable-lm-2)
4. [Stable LM 2 1.6B Technical Report](https://arxiv.org/html/2402.17834v1)
5. [stabilityai/stablelm-2-1_6b · Hugging Face model card](https://huggingface.co/stabilityai/stablelm-2-1%5F6b)
6. [StableLM Review: Stability AI's Open Language Models](https://recatools.com/ai-directory/stablelm/)
7. [Stability AI launches StableLM, an open source ChatGPT alternative — Ars Technica](https://arstechnica.com/information-technology/2023/04/stable-diffusion-for-language-stability-launches-open-source-ai-chatbot/)
8. [StableLM · Hugging Face Transformers docs](https://huggingface.co/docs/transformers/model_doc/stablelm)
---
*Topic: Encyclopedia › Technology and the built world › Computing and digital systems › Modern AI: foundation models, generative AI, and the AI industry › Model families and named models › Large language model families › Open-weight model families*
*Initially written Sep 17, 2026 · Edited: Oct 11, 2026; Oct 15, 2026*
*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*
License: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license