Nemotron 3 Super 120B A12B (Reasoning)
AI model by NVIDIA. Real-time pricing and benchmark data.
Benchmarks
Editorial Analysis
Profile. Nemotron 3 Super 120B A12B (Reasoning) is a budget-tier option from NVIDIA at $0.65/M output tokens. Despite the low price, it scores competitively on Coding Index (37.7) and Intelligence Index (13.6) — making it a candidate for high-volume production workloads where unit cost dominates.
Latency. On latency, Nemotron 3 Super 120B A12B (Reasoning) reports 173 tok/s output speed and 2390ms time-to-first-token. Output speed is in the normal range — most chat and completion applications will be token-bound, not streaming-bound.
Cost model. For a workload of 1M input tokens + 100K output tokens, Nemotron 3 Super 120B A12B (Reasoning) costs $0.26 per million input-equivalent requests. Multiply by your monthly request volume to project the bill, and compare against the speed/correctness profile above to decide if the trade-off is worth it for your workload.
Editorial summary generated from public pricing and benchmark data (Artificial Analysis API). All numbers are sourced — see the Artificial Analysis methodology for benchmark definitions. Last refreshed 2026-10-08.
Compare with similar models
| Model | Input | Output | Speed |
|---|---|---|---|
Nemotron 3 Super 120B A12B (Reasoning)Current | $0.19 | $0.65 | 173 tok/s |
| Granite 4.2 30B | $0.16 | $0.65 | 77 tok/s |
| Qwen3 VL 32B Instruct | $0.16 | $0.64 | 57 tok/s |
| Qwen3 32B (Reasoning) | $0.16 | $0.64 | 106 tok/s |
| Qwen3 32B (Non-reasoning) | $0.16 | $0.64 | 106 tok/s |
| Qwen3 VL 32B (Reasoning) | $0.16 | $0.64 | 88 tok/s |