Loading the catalogue…
NVIDIA
NVIDIA-Nemotron-3-Super-120B-A12B
Released 11 Mar 2026
NVIDIA's open-weight text MoE model with optional reasoning, 12B active parameters and a long context window.
€0.08/€0.36per 1M in / out
Loading the catalogue…
NVIDIA's open-weight text MoE model with optional reasoning, 12B active parameters and a long context window.
| Inference Provider | Sovereignty | Runs in | In / out per 1M | Status | Pin |
|---|---|---|---|---|---|
| L0 | €0.08 / 0.36 | Online |
12 use cases classified by the curator. The note says why.
Native tool-calling support directly matches applications that require structured function invocation and API integration.
Tool-calling support is explicitly documented, making this model well suited to selecting and coordinating external tools in agent workflows.
Tool-calling enables repository and development-tool interaction, while extended thinking supports multi-step coding agents; no coding benchmarks or context-window size are published.
Tool-calling and extended-thinking support the action sequencing required by browser agents, but browser-specific performance and vision capabilities are not documented.
The documented extended-thinking mode supports debugging and causal analysis, though no model-specific software-engineering benchmark evidence is available.
Extended-thinking capability is relevant to analyzing code and reasoning about defects, but the model lacks published SWE-bench or code-review scores and has no specified context window.
Extended thinking can support analytical reasoning over supplied data, but tool execution, code-interpreter support, and data-analysis benchmarks are not documented.
The model's extended-thinking capability supports planning and maintaining coherence in substantial written outputs, but its context window and writing-quality evaluations are unspecified.
Sourced numbers, not our own evaluations. Percentiles are against the whole Stav catalogue.
Computed from aa-intelligence-index, gpqa-diamond, aa-engineering-index, aa-coding-index, livecodebench, terminal-bench-v4, ifbench, aa-lcr.
curl https://api.stav.ai/v1/chat/completions \
-H "Authorization: Bearer $STAV_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "nvidia/nvidia-nemotron-3-super-120b-a12b",
"reasoning_effort": "high",
"messages": [{"role": "user", "content": "Hello!"}]
}'→ X-Stav-Request-Id · X-Stav-Model · X-Stav-Provider → X-Stav-Sovereignty-Level — the rung the request was served at