Groot-19B-A3B-261023

Use it to handle complex reasoning and tool-using workloads, including coding and other long-context tasks, with local or self-hosted inference.

Availability

Weights released in BF16, FP8, NVFP4, and GGUF formats

Pricing model

Not established

Deployment

Local inference and serving with Transformers, vLLM, SGLang, or KTransformers

Privacy

Not established

Licence

Apache-2.0

• The model has about 19B total parameters and activates about 3B per token. ^[1]

  Groot-19B-A3B-261023 has 19B total parameters and activates about 3B per token.

• It supports a native context length of 262,144 tokens and natively supports tool calling. ^[1]

  It has a default context length of **262,144** tokens and natively supports tool calling.

• BF16, FP8, NVFP4, and GGUF (Q4_K_M) weights are released. ^[1]

  **Where it runs.** We release BF16, **FP8**, **NVFP4** and **GGUF (Q4_K_M)** weights.

• The BF16 weights fit on a single 48GB or 80GB GPU. ^[1]

  The BF16 weights (about 39 GB) fit on a single 48GB or 80GB GPU.

• With YaRN scaling, the effective context window can be extended to roughly 1M tokens. ^[1]

  YaRN is built into both vLLM and SGLang, and a scaling factor of 4.0 grows the usable window to roughly 1M tokens.

• The source reports that it beats Qwen3.6-35B-A3B and Gemma-4-26B-A4B on 12 of 16 benchmarks. ^[1]

  **Accuracy.** Groot-19B-A3B-261023 beats both Qwen3.6-35B-A3B and Gemma-4-26B-A4B on 12 of the 16 benchmarks, for example Terminal Bench 2.1 (52.10 vs 43.82 / 26.52), NL2Repo-Bench (35.65 vs 28.77 / 16.55) and LongBench v2 (62.03 vs 48.12 / 47.99).

• A speech model adds multilingual transcription and translation Daily brief 2026-10-26