Quanta is built on a foundation that includes open-source software and openly licensed model weights. This page records those components and the terms under which they are used, as those licenses ask.
The Quanta model family is developed by QuantaCore Labs. Each model's behaviour, alignment, reasoning protocol and Foundational Code of Ethics are QuantaCore Labs' own work. The underlying weights each model builds on are openly licensed and are attributed here.
These run on QuantaCore Labs’ servers. Their weights are not distributed.
| Quanta model | Foundation weights | Copyright | License |
|---|---|---|---|
| Quanta | Qwen3-14B | Alibaba Cloud | Apache 2.0 |
| QuantaCode | Qwen3.8-27B | Alibaba Cloud | Apache 2.0 |
| Quanta-Pro | Qwen3.5-122B-A10B | Alibaba Cloud | Apache 2.0 |
| Quanta-Vision | Qwen3-VL-8B-Instruct | Alibaba Cloud | Apache 2.0 |
| Quanta-Aegis | Mistral‑Small‑24B‑Instruct‑2501 | Mistral AI | Apache 2.0 |
| Quanta Nano | granite‑4.0‑h‑micro | IBM | Apache 2.0 |
| Quanta Nano (images) | Mistral‑Small‑3.1‑24B‑Instruct‑2503 | Mistral AI | Apache 2.0 |
| Quanta Elysium | Qwen‑Image‑2512 | Qwen Team, Alibaba Group | Apache 2.0 |
| Quanta Elysium (4‑step) | Qwen‑Image‑2512‑Lightning | LightX2V | Apache 2.0 |
| Quanta Elysium (prompt reader) | Qwen2.5‑VL‑7B‑Instruct | Qwen Team, Alibaba Group | Apache 2.0 |
| Quanta Kinetic | Wan2.2‑TI2V‑5B | Wan Team, Alibaba Group | Apache 2.0 |
| Quanta Voice | Qwen3‑TTS‑12Hz‑0.6B‑Base | Qwen Team, Alibaba Group | Apache 2.0 |
| Quanta hearing | Qwen3‑ASR‑1.7B | Qwen Team, Alibaba Group | Apache 2.0 |
These weights are downloaded to your device and run in your browser, so their licenses apply to the copy you receive.
| Quanta model | Foundation weights | Copyright | License |
|---|---|---|---|
| Quanta‑Micro‑0.8B‑v4d | Qwen3.5‑0.8B | Alibaba Cloud | Apache 2.0 |
| Quanta‑Balanced‑1.5B‑v3 | Qwen2.5‑1.5B‑Instruct | Alibaba Cloud | Apache 2.0 |
| Quanta‑Max‑2B‑v3 | Qwen3.5‑2B | Alibaba Cloud | Apache 2.0 |
| Quanta‑Macro‑9B | Qwen3.5‑9B | Alibaba Cloud | Apache 2.0 |
| Speech input | whisper‑tiny.en | OpenAI | MIT |
The foundation weights listed above are licensed under the Apache License, Version 2.0 — with the exception of whisper-tiny.en, which is licensed under the MIT License reproduced below. You may obtain a copy of the Apache License at apache.org/licenses/LICENSE-2.0.
In accordance with Section 4(b) of the Apache License: Quanta, QuantaCode, Quanta-Balanced-1.5B-v3, Quanta-Max-2B-v3, Quanta-Macro-9B and Quanta Voice are modified works. QuantaCore Labs fine-tuned their foundation weights. The other models run their foundation weights as published, under QuantaCore Labs’ own instruction, safety screening and inference configuration.
whisper-tiny.en, used for on-device speech input, is licensed as follows.
| Component | Used for | License |
|---|---|---|
| Ollama | Model serving runtime | MIT |
| Geist | Base of the “Quanta Geneva” text system (self-hosted) | SIL Open Font License 1.1 |
| Geist Mono | Base of “Quanta Geneva Mono” — code blocks & editors (self-hosted) | SIL Open Font License 1.1 |
| Space Grotesk | Display typeface | SIL Open Font License 1.1 |
| Inter | Body typeface | SIL Open Font License 1.1 |
| JetBrains Mono | Monospace typeface | SIL Open Font License 1.1 |
Quanta, QuantaCode, Quanta-Vision and QuantaCore Labs are marks of QuantaCore Labs. All other names and marks referenced on this page belong to their respective owners, and their inclusion here is attribution required by the licenses above — not an endorsement, affiliation or partnership of any kind.