Compare models. Find the differences.
Add, remove, or change models in the table.
Benchmark snapshot
0 benchmarks in common.
4 models selected. 0 benchmarks in common.
Reported scores; test settings may differ.
| Benchmark | NVIDIA | OpenAI | Anthropic | Google |
|---|---|---|---|---|
| Arena ELOHuman preference | 1,427 | Not in our dataset | 1,456 | 1,438 |
| LiveBench 1.0General reasoning | 67.4% | Not in our dataset | Not in our dataset | Not in our dataset |
| GPQA DiamondScientific knowledge | Not in our dataset | Not in our dataset | 83.4% | 82.3% |
| HLEAcademic reasoning | 26.7% | Not in our dataset | 17.7% | 8.7% |
| MMLU ProMulti-task understanding | 86.8% | Not in our dataset | 87.5% | 82.6% |
| SimpleBenchCommonsense reasoning | 41.7% | Not in our dataset | 54.3% | Not in our dataset |
| AIME 2025Mathematics | Not in our dataset | Not in our dataset | 87.0% | Not in our dataset |
| GSOGraduate science | Not in our dataset | Not in our dataset | 14.7% | Not in our dataset |
| SWE-Bench VerifiedAgentic coding | 71.9% | 74.5% | 77.2% | Not in our dataset |
| LiveCodeBenchCompetitive coding | Not in our dataset | Not in our dataset | Not in our dataset | 77.1% |
| Tau-BenchAgentic tool use | 86.4% | Not in our dataset | 86.2% | Not in our dataset |
| Terminal-BenchAgentic terminal | Not in our dataset | 43.4% | 42.8% | Not in our dataset |
| ARC-AGI-2Visual reasoning | Not in our dataset | Not in our dataset | 13.6% | Not in our dataset |
Pricing & context Published USD rates per 1M tokens
| Specification | Nemotron 3 Ultra 550B-A55B | GPT-5 Codex Non-thinking | Claude Sonnet 4 4.5 Thinking | Gemma 4 26B A4B IT Thinking |
|---|---|---|---|---|
| Input | $0.6 | $0.625 | $3 | $0.095 |
| Output | $2.4 | $5 | $15 | $0.284 |
| Context window | 262K | 400K | 1.0M | 262K |
| Pricing details | Explore pricing → | Explore pricing → | Explore pricing → | Explore pricing → |
How to read this comparison
Each column keeps the exact model version you selected. Shared benchmarks are the intersection of their reported results; “all coverage” also shows where evidence is missing. Missing results never count as zero.
These are published benchmark scores, not a controlled head-to-head test. The available evidence does not establish matching evaluation settings, so we show no cross-model winner highlights or deltas. Composite quality scores are omitted because they can summarize different benchmark sets.
Scores and statistics are computed upstream. This page reads a fixed data snapshot; its date describes the snapshot, not the date of every measurement.