AgenticaApr 8, 2025
DeepCoder 14B
Agentica and Together AI's 14B code-reasoning model trained with reinforcement learning, reporting coding scores close to OpenAI o3-mini.
- Type
- Code
- Context
- 131K
- Weights
- Open
Tag
Agentica on The Latent Daily: 2 models. News, explainers and guides on Agentica, each linked to its sources.
Agentica and Together AI's 14B code-reasoning model trained with reinforcement learning, reporting coding scores close to OpenAI o3-mini.
A 1.5B reasoning model from Agentica trained with reinforcement learning, which matched much larger models on competition math at release.