Neutron 1.2
A language model that works in your language, not around it.
নিউট্রন আপনার ভাষাতেই কাজ করে — আলাদা কোনো সেটিংস বা অনুবাদের ধাপ ছাড়াই।
Built by Respone · Dhaka, Bangladesh
What Neutron is
Neutron reads and writes, reasons through problems, works with code, and understands documents and images — in whichever language you write in, with no configuration and no translation step in between.
| Version | Neutron 1.2 |
| Released | 8 September 2026 |
| Inputs | text · code · images · documents |
| Output | text |
| Languages | broad multilingual coverage, evaluated below in five |
| Licence | Proprietary — Neutron Model Licence |
Benchmarks
Every number below was measured by us on 8 September 2026, in one harness, with byte-identical prompts, the same answer parser, and the same sampled items for every model. Nothing here is copied from a vendor's published figures.
The same numbers as text
| Model | Global-MMLU বাংলা |
Belebele বাংলা |
Belebele English |
Belebele हिन्दी |
Belebele العربية |
Belebele Español |
Mean |
|---|---|---|---|---|---|---|---|
| Neutron 1.2 | 62.0 | 75.0 | 97.0 | 82.0 | 94.0 | 92.0 | 83.7 |
| Gemma 4 31B | 79.0 | 89.0 | 96.0 | 87.0 | 92.0 | 93.0 | 89.3 |
| Qwen3.8-27B | 59.0 | 73.0 | 97.0 | 73.0 | 90.0 | 91.0 | 80.5 |
| GPT-5 nano | 32.0 | 43.0 | 92.0 | 45.0 | 68.0 | 79.0 | 59.8 |
Where Neutron competes
Neutron is priced and sized against the small, fast tier — the models people actually reach for when cost matters. Against that tier it is not close:
+24 points on average, and +30 and +32 points in Bengali — the two tasks where a cheap model is most likely to be deployed in Bangladesh and least likely to hold up.
What the numbers also say
Gemma 4 31B is ahead of Neutron in Bengali — 79 vs 62 on Global-MMLU and 89 vs 75 on Belebele. Both gaps exceed the confidence interval, so they are real, not noise. We publish this because a benchmark table that only contains results favourable to the publisher is not a benchmark table.
Neutron matches or beats it in English (97 vs 96) and Arabic (94 vs 92), and Bengali is where our next release is aimed.
Method
- 100 items per task, sampled with a fixed seed from the public test splits of Global-MMLU and Belebele.
- Identical prompts for every model, built once and shared.
- One parser for every model, deliberately lenient — it accepts a bare
letter,
The answer is B, orB) …— so that models which wrap the answer in prose are not penalised for formatting. - Zero-shot, temperature 0, no chain-of-thought prompting, no few-shot examples, no per-model prompt tuning.
- Comparison models were served through OpenRouter; Neutron ran on our own hardware at 4-bit NF4, a handicap relative to the higher-precision serving the hosted models receive.
- 95% CI ≈ ±10 points at n=100. Differences under 10 points are indicative, not conclusive. We would rather state that than imply a precision the sample size does not support.
Larger runs are in preparation and will be published here with the same disclosure.
Training
Neutron 1.2 is a supervised fine-tune over a permissively licensed open-weight foundation, trained by Respone on a mixture of identity, agentic, coding and multilingual data.
| Training loss | 0.8065 → 0.4959 |
| Held-out evaluation loss | 0.5099 |
| Held-out token accuracy | 83.09% |
| Optimisation steps | 255 (3 epochs) |
| Examples carrying reasoning traces | 83% |
The 0.014 gap between evaluation and training loss indicates the model generalised rather than memorised its training set.
Behaviour
Verbatim, unedited responses from Neutron 1.2.
তুমি কে?
আমি Neutron — Respone-এর তৈরি একটি ভাষা মডেল। লেখালেখি, বিশ্লেষণ, কোডিং আর এজেন্টিক কাজে সাহায্য করি, যেকোনো ভাষায়। কী নিয়ে কাজ করছেন?
Who are you?
I'm Neutron, a language model built by Respone. I work across text, code, documents and images, in whichever language you prefer. What are you working on?
What are you bad at? Be honest.
I can state wrong facts confidently, especially on niche topics, dates and figures — verify anything that matters. Long arithmetic and precise counting are unreliable; give me a tool for those. In less widely written languages my fluency can outrun my accuracy, which is a bad combination.
Limitations
- It can be confidently wrong. Verify names, dates, figures and citations.
- Knowledge has a cutoff. Recent events may be missing or wrong.
- Arithmetic and counting are unreliable. Give it a tool.
- No audio input. Voice interfaces transcribe first, so tone is lost before the model sees anything.
- Bengali is not yet our strongest result. The table above says so plainly; it is the target of the next release.
Access
API — an OpenAI-compatible endpoint. Point an existing client at a different base URL; no SDK to learn, no rewrite.
Private deployment — Neutron inside your own infrastructure, fully offline-capable, for work that cannot leave your premises.
Evaluation — bring a real task from your own work. It is the only test that tells you anything useful.
Contact hello@respone.ai.
Licence and use
Neutron is proprietary software. It is not open-source, not open-weight, and not available for download.
This repository contains no model. It holds this card and its figures, nothing else. There is no checkpoint here to clone, pull or quantise.
The weights are released only under a written agreement, and the Neutron Model Licence governs them:
- no copying, hosting, mirroring, resale or redistribution of the weights;
- no extraction or publication of weights, adapters or training data;
- no reverse-engineering of the architecture, parameters or training corpus;
- no training or distilling a competing model on Neutron's output;
- Neutron and Respone are trademarks — no right to use the names is granted with the model.
What is free: this card. Read it, quote it, cite it, benchmark against the numbers in it. Naming Neutron in a comparison or review needs no permission from anyone.
Licensing enquiries: hello@respone.ai
Team
| Masud Ashraf Taha | Founder & Chief Executive Officer |
| Sharan Sifat | Founder & Chief Technology Officer — model development, training and evaluation |
About Respone
Respone is a Bangladeshi company building language technology that treats Bengali as a first-class language rather than an afterthought.
Neutron is built on open-source foundation models under permissive licences. The continued training, language adaptation, instruction tuning, safety alignment, evaluation and deployment engineering are ours. Technical foundations and the third-party component list are documented and available under NDA, and are listed in the Third-Party Components schedule of our service agreement.