{"id":"laghu","slug":"laghu","name":"Laghu","orgId":"udgam-labs","family":"laghu","releaseDate":null,"availability":"announced","openWeights":true,"license":null,"paramsTotal":null,"paramsSource":null,"contextWindow":null,"knowledgeCutoff":null,"modelCardUrl":"https://udgamlabs.com/blog/laghu","hfRepo":null,"modalities":["text"],"aliases":["Laghu","laghu","udgamlabs/Laghu","Udgam Laghu"],"aliasNote":null,"origin":"tracked","tracked":{"trackedSince":"2026-10-24","announced":"2026-10-22","blurb":"Udgam Labs' 152M-parameter English chat model with a thinking on/off switch, trained from scratch with a $146 main pretraining run and released fully open: weights, late checkpoints, fine-tuning data, evaluation outputs and code. The 163 MB model runs in the browser. Apache-2.0.","api":"none","openWeights":"yes","params":{"text":"152M (151.8M; 28 layers, width 640)","source":"Udgam Labs model page"},"facts":[{"label":"Architecture","value":"Gemma 4 text architecture: 10 query and 2 key-value heads; three 512-token sliding-window layers then one global layer, repeated"},{"label":"Training","value":"Trained from scratch with Udgam's own 32,768-token tokenizer: about 70B tokens read in pretraining at 2K, a 1B-token stretch at 32K, then fine-tuning on 41,467 conversations"},{"label":"Cost","value":"Main pretraining run $146.26 (8× A100 80GB, 13.2 hours); about $194 for the whole project"},{"label":"Context","value":"16K tokens for use (stretched at 32K; recall fails beyond about 16K)"},{"label":"Modes","value":"Thinking on or off; Udgam recommends off except for short arithmetic"},{"label":"Release","value":"Laghu, Laghu-Base, Laghu-Base-32K, Laghu-Checkpoints (late training states only; earlier checkpoints were lost), GGUF and MLX builds, the laghu-sft and laghu-evals datasets, and code on GitHub (udgamlabs/laghu)"},{"label":"License","value":"Apache-2.0 for weights and code; nanochat-derived files keep their MIT notice"}],"claims":["Lab claim: level with SmolLM2-135M-Instruct on judged chat quality (51.0% win rate) and multi-turn follow-ups (41.7% vs 37.5%); behind Falcon-H1-Tiny-90M-Instruct on all three of Udgam's pre-registered chat tests, so it missed Udgam's own release bar.","Lab claim: GSM8K 7.9% with thinking on, IFEval prompt-level strict 26.1.","Lab claim: Laghu-Base scores 0.919 bits per byte on text published in September–October 2026, against 0.941 for SmolLM2-135M and 0.979 for Falcon-H1-Tiny-90M-Base."],"sources":[{"kind":"lab","title":"Laghu release post (Udgam Labs blog)","url":"https://udgamlabs.com/blog/laghu","date":"2026-10-22"},{"kind":"lab","title":"Laghu model page (Udgam Labs)","url":"https://udgamlabs.com/models/laghu","date":"2026-10-22"},{"kind":"repo","title":"udgamlabs/Laghu model card (Hugging Face)","url":"https://huggingface.co/udgamlabs/Laghu"},{"kind":"repo","title":"udgamlabs/laghu (GitHub)","url":"https://github.com/udgamlabs/laghu"},{"kind":"lab","title":"Udgam Labs on X","url":"https://x.com/udgamlabs/status/2113272462062637075","date":"2026-10-22"}],"firstBenchmark":null},"organization":{"id":"udgam-labs","name":"Udgam Labs","slug":"udgam-labs","country":null,"url":"https://udgamlabs.com","color":"#11110F","mark":"U"},"price":null,"answer":{"word":"Pending benchmarks.","scope":"Announced Oct 22, 2026; no independent results yet.","lead":"As of Nov 3, 2026, Laghu is pending benchmarks on isitsota: announced Oct 22, 2026, with no independent results on the boards we track yet.","confidence":"low"},"status":"PENDING_BENCHMARKS","verdicts":[]}