Compare / Trinity Large Thinking vs Nemotron 3 Super
Trinity Large Thinking vs Nemotron 3 Super
Which is cheaper, which is more capable, and which does better at each kind of work, from 27 results on 3 sources that measured both models.
Trinity Large Thinking or Nemotron 3 Super? The short answer
- Nemotron 3 Super is cheaper: $0.16 against $0.39 per million tokens, 2.4 times less.
- Nemotron 3 Super scores higher on the Artificial Analysis Intelligence Index: 12.8 against 10.8.
- Trinity Large Thinking writes faster: 344 tokens a second against 150.
- On our business benchmarks, Trinity Large Thinking ranks higher for writing.
- On our business benchmarks, Nemotron 3 Super ranks higher for agents and tool use, reasoning and knowledge and sticking to the facts.
At a glance
| Trinity Large Thinking | Nemotron 3 Super | |
|---|---|---|
| Price per million tokens | $0.25 in · $0.80 out | $0.085 in · $0.40 out |
| Blended (3 in : 1 out) | $0.39 | $0.16 |
| Cheapest host, blended | – | $0.16 (DeepInfra, BF16) |
| Intelligence Index | 10.8 | 12.8 |
| Spring Prompt overall | – | – |
| Output speed, tokens a second | 344 | 150 |
| Context window | 262,144 tokens | 262,144 tokens |
| Weights | Open | Open |
| Developer based in | the US | the US |
Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.
Which is better for what
Each model's rank among models available today, for each kind of work, from our own benchmarks and licensed sources. The better rank is in green.
| Use case | Trinity Large Thinking | Nemotron 3 Super |
|---|---|---|
| Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls | 116 of 161 | 110 of 161 |
| Reasoning and knowledgeHard questions across science, maths and general knowledge | 120 of 195 | 103 of 195 |
| WritingWhat people prefer in blind comparisons, and judged writing quality | 129 of 187 | 135 of 187 |
| Sticking to the factsSummarising without inventing things, and factual answers | 142 of 153 | 137 of 153 |
Every shared result
| Benchmark | Metric | Trinity Large Thinking | Nemotron 3 Super |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,371 | 1,390 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,367 | 1,350 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,332 | 1,303 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,401 | 1,393 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,357 | 1,343 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,368 | 1,360 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,343 | 1,325 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 25.8 | 37.7 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 10.8 | 12.8 |
| Artificial Analysisreported | GPQA Diamond% of questions | 75.2% | 80.0% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 15.8% | 20.8% |
| Artificial Analysisreported | IFBench% of instructions | 56.3% | 71.5% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 38.0% | 65.7% |
| Artificial Analysisreported | Output speedtokens per second | 344 | 150 |
| Artificial Analysisreported | SciCode% of problems | 40.6% | 36.2% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 20.6% | 38.6% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 0.5% | 0.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 22.7% | 28.8% |
| Artificial Analysisreported | Time to first answer tokenseconds | 6.88 | 14.1 |
| Artificial Analysisreported | Τ-bench banking% of tasks | 5.8% | 10.3% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 90.1% | 67.8% |
| OpenHands Indexreported | Average% resolved | 32.1% | 36.2% |
| OpenHands Indexreported | Front end (SWE-Bench Multimodal)% resolved | 25.0% | 20.6% |
| OpenHands Indexreported | Greenfield (Commit0)% resolved | 12.5% | 12.5% |
| OpenHands Indexreported | Information gathering (GAIA)% resolved | 32.7% | 40.0% |
| OpenHands Indexreported | Issue resolution (SWE-Bench)% resolved | 56.8% | 62.0% |
| OpenHands Indexreported | Testing (SWT-Bench)% resolved | 33.3% | 45.7% |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.
Quick answers
Is Trinity Large Thinking better than Nemotron 3 Super?
It depends on the task. Among models available today, on our benchmarks and licensed sources, Trinity Large Thinking ranks higher for writing; Nemotron 3 Super ranks higher for agents and tool use, reasoning and knowledge and sticking to the facts.
Which is cheaper, Trinity Large Thinking or Nemotron 3 Super?
Nemotron 3 Super costs $0.16 per million tokens (three input to one output), against $0.39 for Trinity Large Thinking.
Which is faster, Trinity Large Thinking or Nemotron 3 Super?
Trinity Large Thinking writes about 344 tokens a second on its usual API, against 150 for Nemotron 3 Super (Artificial Analysis).
Run Trinity Large Thinking and Nemotron 3 Super on your own prompt
Benchmarks aren't your data. Try both side by side in the playground, with the cost of every answer.