Nous ResearchNousCoder
NousCoder 14B
NousCoder 14B applies reinforcement learning to Qwen3-14B on verifiable programming problems.
Training and model behavior
The release used 24,000 coding problems and 48 B200 GPUs over four days. Its reported LiveCodeBench result is for a specific version and date range. The weights use an extended YaRN configuration.NousCoder 14B model card
Working with this model
Practical guidance. Evaluate with problems outside the training period. Record compilation failures, execution results and time limits. Competitive-programming accuracy does not establish repository-editing performance.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | NousResearch/NousCoder-14B |
| Context window | 81,920 tokens Configured position limit; practical serving limits depend on the runtime. NousCoder 14B configuration |
| Base model | Qwen/Qwen3-14BNousCoder 14B model card |
| Repository license label | apache-2.0NousCoder 14B model card |