Trendingby GlyphSmall language models
Rising · emerged June 2023 · high October 2026 · -2% on the month before · 20 mentions on Trending in 14 days
Compact models, a few billion parameters or less, trained to be good enough and cheap to run anywhere. Named by Gunasekar et al., Microsoft Research: Textbooks Are All You Need (phi-1) (Jun 20, 2023). Also called SLM, SLMs, Small models, Tiny models.
Index 0 to 100, monthly
Signals
Hacker News stories, last 12 months57 stories, +0% on the 12 months before
arXiv papers, last 12 months697 papers, +53% on the 12 months before
Wikipedia views, October 20264,188, +85% on a year beforeProduct releases naming it1 since Feb 19, 2026GitHub repositories2,916| Signal | Reads as | |
|---|---|---|
| Hacker News stories, last 12 months57 stories, +0% on the 12 months before | 57 stories, +0% on the 12 months before | Adoption |
| arXiv papers, last 12 months697 papers, +53% on the 12 months before | 697 papers, +53% on the 12 months before | Adoption |
| Wikipedia views, October 20264,188, +85% on a year before | 4,188, +85% on a year before | Adoption |
| Product releases naming it1 since Feb 19, 2026 | 1 since Feb 19, 2026 | Adoption |
| GitHub repositories2,916 | 2,916 | Adoption |
Companies
| Name | Mentions |
|---|---|
| 1 |
Releases
1 release since Feb 19, 2026| Name | |
|---|---|
| May 18, 20265 months ago |
Models
Phi 4MicrosoftPhi 4 PlusMicrosoftGemma 3 1bGoogleGemma 3 4bGoogleQwen3 4bAlibabaQwen3.5 4BAlibabaLlama 3.2 3bMeta
| Name |
|---|
| Phi 4Microsoft |
| Phi 4 PlusMicrosoft |
| Gemma 3 1bGoogle |
| Gemma 3 4bGoogle |
| Qwen3 4bAlibaba |
| Qwen3.5 4BAlibaba |
| Llama 3.2 3bMeta |
Papers
| Name | |
|---|---|
| Nov 24, 202511 months ago | |
| Nov 20, 20241 year ago |
GitHub projects
2,916 repositoriesmicrosoft/PhiCookBookThis is a Phi Family of SLMs book for getting started with Phi Models. Phi a family of open sourced AI models developed by Microsoft. Phi mokpu/kenlmKenLM: Faster and Smaller Language Model Queriessammyuri/craftgptSmall language model built in Minecraft.mbzuai-oryx/MobiLlama[ICLR-2025-SLLM Spotlight 🔥]MobiLlama : Small Language Model tailored for edge devicesUcas-HaoranWei/Vary-toyOfficial code implementation of Vary-toy (Small Language Model Meets with Reinforced Vision Vocabulary)HKUDS/LightReasoner[ACL 2026 Oral] "LightReasoner: Can Small Language Models Teach Large Language Models Reasoning?"
| Name | Stars | |
|---|---|---|
| microsoft/PhiCookBookThis is a Phi Family of SLMs book for getting started with Phi Models. Phi a family of open sourced AI models developed by Microsoft. Phi mo · May 7, 2024 2 years ago | May 7, 20242 years ago | 3,888 |
| kpu/kenlmKenLM: Faster and Smaller Language Model Queries · Nov 15, 2011 14 years ago | Nov 15, 201114 years ago | 2,807 |
| sammyuri/craftgptSmall language model built in Minecraft. · Sep 28, 2025 1 year ago | Sep 28, 20251 year ago | 726 |
| mbzuai-oryx/MobiLlama[ICLR-2025-SLLM Spotlight 🔥]MobiLlama : Small Language Model tailored for edge devices · Feb 23, 2024 2 years ago | Feb 23, 20242 years ago | 671 |
| Ucas-HaoranWei/Vary-toyOfficial code implementation of Vary-toy (Small Language Model Meets with Reinforced Vision Vocabulary) · Jan 10, 2024 2 years ago | Jan 10, 20242 years ago | 630 |
| HKUDS/LightReasoner[ACL 2026 Oral] "LightReasoner: Can Small Language Models Teach Large Language Models Reasoning?" · Sep 29, 2025 1 year ago | Sep 29, 20251 year ago | 610 |
Related movements
Local AIRunning language and image models on your own laptop, phone or server instead of a cloud API.Open-weight modelsModels whose trained weights anyone can download, run and fine-tune: Llama, Qwen, DeepSeek, Mistral and more.Synthetic dataTraining and evaluation data generated by models or simulators instead of collected from people.
| Movement | Stage |
|---|---|
| Local AIRunning language and image models on your own laptop, phone or server instead of a cloud API. | Peak |
| Open-weight modelsModels whose trained weights anyone can download, run and fine-tune: Llama, Qwen, DeepSeek, Mistral and more. | Rising |
| Synthetic dataTraining and evaluation data generated by models or simulators instead of collected from people. | Plateau |
History
Stage changed from peak to rising
Tracking started: emerged June 2023, peak, 4 other names recorded
| Date | What changed |
|---|---|
| Nov 1, 2026 | Stage changed from peak to rising |
| Sep 24, 2026 | Tracking started: emerged June 2023, peak, 4 other names recorded |
Sources: Curve: Hacker News story titles, Wikipedia pageviews and arXiv papers by month, refreshed Nov 1, 2026. Related entities from the other Glyph Ibis sources' public APIs, matched by name. How stages work.