Jamba Reasoning 3B
AI21 LabsOpen weights
Jamba Reasoning 3B is an open-weights model from AI21 Labs. No provider in the open price lists serves it yet.
- Maker
- AI21 Labs
- Licence
- Open weights
- Context window
- —
- Providers listed
- —
- Lowest input, per 1M
- —
- Lowest output, per 1M
- —
Providers serving Jamba Reasoning 3B
No provider in the open price lists serves this model yet.
Benchmark scores
Published scores only, each linked to its source. Compare every model in Bench rankings
| Benchmark | Score |
|---|---|
GPQA Diamond PhD-level scientific reasoning, pass@1, no tools | 33.3% |
Humanity's Last Exam (text-only) Reasoning & knowledge, no tools | 3.8% |
τ²-Bench Telecom Agentic tool use (telecom), with a Qwen3 model as the simulated user | 15.8% |
LiveCodeBench Competitive programming, Artificial Analysis's problem set (dates not published) | 21% |
IFBench Instruction following, 294 prompts, 5 runs each | 52.4% |
Terminal-Bench Hard Agentic coding & terminal use, Artificial Analysis's 44-task hard subset | 0.8% |
LCR Long-context reasoning, AA-LCR v1.1 | 6.3% |
Omni Accuracy Omniscience knowledge accuracy | 6.7% |
Non-Hallucination 1 − hallucination rate | 31% |