NVIDIALlama 3.3 Nemotron Super 49B v1.5
Llama 3.3 Nemotron Super 49B v1.5 is a NVIDIA open-weights model, released 18 Mar 2025. Its best configuration (thinking reasoning effort) ranks #629 of 785 on the BenchLeader Index at 41.5 ±2.7, in the lower half. The ± is the point: 278 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (41) and lowest in long context (33). At $0.400 per million tokens blended it is cheaper than most ranked models. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 18 Mar 2025.stale: no new result in six months
- Blended price
- $0.400/M
- $0.400 in · $0.400 out
- Output speed
- –
- First answer
- –
- Context
- 131k
- Full answer
- –
- Cost per run
- –
- Released
- 18 Mar 2025
- Overall index42
- Reasoning41
- Agents & tools34
- Knowledge35
- Instruction following41
- Long context33
- Composite39
Versions
NVIDIA has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| Llama Nemotron Super 49B v1.5thinking | 25 Jul 2025 | 43.6 | #565 |
| Llama 3.3 Nemotron Super 49B v1.5thinkingthis page | 18 Mar 2025 | 41.5 | #629 |
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 41.4 | #632 | – | – | $0.0003 | Agents & tools 34 · Composite 38 · Instruction following 42 · Knowledge 39 · Long context 30 · Reasoning 41 |
| thinkingbest | 41.5 | #629 | – | – | $0.0003 | Agents & tools 34 · Composite 39 · Instruction following 41 · Knowledge 35 · Long context 33 · Reasoning 41 |
| not stated | – | – | – | – | $0.0003 | Coding 53 · Human preference 51 · Maths 56 · Reasoning 51 |
How its index has moved
9 Sept 2026 to 21 Oct 2026- thinking
- no reasoning
The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| Arena Hard Promptseffort not stated | – | – | 1364#203 | Arena (LMArena) | |
| GPQA Diamond (AA)not in index | 51.7%#417 | 64.3%#348 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 3.8%#563 | 5.6%#420 | – | Artificial Analysis | |
| CritPt | 0.0%#315 | 0.0%#315 | – | Artificial Analysis |
Coding
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Coding | 1403#191 | Arena (LMArena) |
Agents & tools
| Benchmark | no reasoning | thinking | Source | Trend |
|---|---|---|---|---|
| Terminal-Bench Hard | 0.0%#362 | 0.0%#362 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | – | 26.9%#279 | Artificial Analysis |
Maths
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Maths | 1387#166 | Arena (LMArena) |
Knowledge
| Benchmark | no reasoning | thinking | Source | Trend |
|---|---|---|---|---|
| AA-Omniscience | -50.3#387 | -58.3#452 | Artificial Analysis | |
| AA-Omniscience: accuracynot in index | 13.5%#455 | 17.1%#373 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 26.3%#235 | 9.1%#440 | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| IFBench | 39.5%#268 | 38.1%#282 | – | Artificial Analysis | |
| Arena Instruction Followingnot in indexeffort not stated | – | – | 1325#212 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1343#209 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1305#211 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1340#205 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1343#210 | Arena (LMArena) |
Long context
| Benchmark | no reasoning | thinking | Source | Trend |
|---|---|---|---|---|
| AA-LCR | 12.0%#462 | 18.3%#443 | Artificial Analysis |
Composite
| Benchmark | no reasoning | thinking | Source | Trend |
|---|---|---|---|---|
| AA Intelligence Index v4.3.2 | 7.3#485 | 8.9#421 | Artificial Analysis |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0003 | – | – |
| Summarize a 30-page report | 12,000 / 600 | $0.0050 | – | – |
| Code edit | 6,000 / 1,500 | $0.0030 | – | – |
| Agentic coding session | 60,000 / 4,000 | $0.026 | – | – |
| Structured extraction | 2,000 / 200 | $0.0009 | – | – |
More from NVIDIA
See also
Questions people ask
- What are Llama 3.3 Nemotron Super 49B v1.5's benchmark scores?
- As of 25 Oct 2026, Llama 3.3 Nemotron Super 49B v1.5 (thinking effort) scores 41.5 ±2.7 on the BenchLeader Index, strongest in instruction following (41) and weakest in long context (33). That rests on 6 independent benchmark results, and every score on this page links to its publisher.
- How does Llama 3.3 Nemotron Super 49B v1.5 rank?
- Llama 3.3 Nemotron Super 49B v1.5 (thinking effort) is #629 of 785 ranked configurations on 25 Oct 2026. 278 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Llama 3.3 Nemotron Super 49B v1.5 cost per 1M tokens?
- As of 25 Oct 2026, Llama 3.3 Nemotron Super 49B v1.5 costs $0.400 per 1M input tokens and $0.400 per 1M output tokens, $0.400 blended (three input tokens per output token).
- When was Llama 3.3 Nemotron Super 49B v1.5 released?
- NVIDIA released Llama 3.3 Nemotron Super 49B v1.5 on 18 Mar 2025. BenchLeader has results for it at 3 reasoning-effort settings.
Data as of 25 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Llama 3.3 Nemotron Super 49B v1.5: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/llama-3-3-nemotron-super-49b.