EssentialAI: Rnj 1 Instruct
Essential AIID: essentialai/rnj-1-instruct
Rnj-1 is an 8B-parameter, dense, open-weight model family developed by Essential AI and trained from scratch with a focus on programming, math, and scientific reasoning. The model demonstrates strong performance across multiple programming languages, tool-use workflows, and agentic execution environments (e.g., mini-SWE-agent).
Pricing per 1M Tokens
| Input (Prompt) | $0.15 |
| Output (Completion) | $0.15 |
| Cache Read | Free |
| Cache Write | Free |
| Image | N/A |
Specifications
| Context Length | 33K |
| Max Output Tokens | N/A |
| Input Modalities | Text |
| Output Modalities | Text |
| Tokenizer | Other |
| Instruct Type | N/A |
| Top Provider Context | 33K |
| Top Provider Max Output | N/A |
| Moderated | No |
Compare this model
See how EssentialAI: Rnj 1 Instruct stacks up against other models.
Last updated: March 23, 2026
First tracked: March 23, 2026