NVIDIALlama 3.3 Nemotron Super 49B v1.5
Best configuration ranks #408 of 610 on the BenchLeader Index at 44.7 ±5.6. Last measured 2 Sept 2026. Released 25 Jul 2025.
- Blended price
- $0.400/M
- $0.400 in · $0.400 out
- Output speed
- –
- First answer
- –
- Context
- 131k
- Overall index45
- Reasoning52
- Coding53
- Agents & tools34
- Knowledge40
- Instruction following42
- Human preference51
- Long context31
- Composite38
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. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| thinking | 41.9 | #476 | – | – | – | Agents & tools 34 · Composite 40 · Instruction following 41 · Knowledge 36 · Long context 34 |
| defaultbest | 44.7 | #408 | – | – | $0.0003 | Agents & tools 34 · Coding 53 · Composite 38 · Human preference 51 · Instruction following 42 · Knowledge 40 · Long context 31 · Reasoning 52 |
Benchmark results
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 | thinking | default | Source |
|---|---|---|---|
| LMArena Hard Prompts | – | 1362#182 | LMArena |
| GPQA Diamond (AA)not in index | 64.3%#332 | 51.7%#395 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 5.6%#345 | 3.8%#478 | Artificial Analysis |
Coding
| Benchmark | thinking | default | Source |
|---|---|---|---|
| LMArena Coding | – | 1403#170 | LMArena |
Agents & tools
| Benchmark | thinking | default | Source |
|---|---|---|---|
| Terminal-Bench Hard | 0.0%#345 | 0.0%#345 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 26.9%#270 | – | Artificial Analysis |
Knowledge
| Benchmark | thinking | default | Source |
|---|---|---|---|
| AA-Omniscience | -58.3#376 | -50.3#314 | Artificial Analysis |
Instruction following
| Benchmark | thinking | default | Source |
|---|---|---|---|
| IFBench | 38.1%#269 | 39.5%#258 | Artificial Analysis |
Human preference
| Benchmark | thinking | default | Source |
|---|---|---|---|
| LMArena Text | – | 1343#187 | LMArena |
Long context
| Benchmark | thinking | default | Source |
|---|---|---|---|
| AA-LCR | 18.3%#370 | 12.0%#389 | Artificial Analysis |
Composite
| Benchmark | thinking | default | Source |
|---|---|---|---|
| AA Intelligence Index | 8.9#346 | 7.3#406 | 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 | – | – |
| Summarise 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 | – | – |
See also
Data as of 9 Sept 2026. Compare these configurations.