Opus Research

Independent research on small language models, done entirely on consumer hardware. We publish weights, methodology, and the experiments that failed.

Pretraining

• opus-1.5 — 0.88B model pretrained from scratch in 42 hours on 2× RTX 4090.
• opus-2.0 — successor, currently paused on compute, not on research.

Fine-tunes

• bernard-qwen3-14b-thinking — can fine-tuning change a model's reasoning, not just its voice? Yes, and the card documents what that costs.
• bernard-gpt-oss-20b-lora — style-transfer LoRA where the worst validation loss was the right checkpoint to ship.
• gemma-2-2b-thinking — adding chain-of-thought to a base model that lacks it.

Classifiers

• opus-moderation-1 — toxicity + jailbreak detection in one 149M model. The card shows three failed recipes and the identity bias each one had.
• opus-emotion-1 — 28 emotions, trained in 14 minutes on an unsupported AMD GPU.

Datasets

• opus-thinking-10k — ~9k chain-of-thought examples.

Team

• spoodyzz (Michael)
• cgcristi0 (CGCristi)
• targeteater (target)

Our constraint is compute, not ideas. Open to collaboration and grant programs — reach us through the organization page.