Compare models. Find the differences.
Add, remove, or change models in the table.
Benchmark snapshot
4 benchmarks in common.
4 models selected. 4 benchmarks in common.
Reported scores; test settings may differ.
| Benchmark | OpenAI | Qwen | Qwen | DeepSeek |
|---|---|---|---|---|
| Arena ELOHuman preference | 1,435 | Not in our dataset | 1,416 | 1,422 |
| LiveBench 1.0General reasoning | Not in our dataset | 64.0% | Not in our dataset | Not in our dataset |
| GPQA DiamondScientific knowledge | 85.7% | 87.8% | 86.6% | 82.4% |
| HLEAcademic reasoning | 24.8% | 24.0% | 25.3% | 25.1% |
| MMLU ProMulti-task understanding | 87.1% | 86.2% | 86.7% | 85.0% |
| SimpleBenchCommonsense reasoning | 56.7% | Not in our dataset | Not in our dataset | Not in our dataset |
| AIME 2025Mathematics | 94.6% | Not in our dataset | Not in our dataset | 93.1% |
| GSOGraduate science | 6.9% | Not in our dataset | Not in our dataset | Not in our dataset |
| AiderCode editing | 88.0% | Not in our dataset | Not in our dataset | Not in our dataset |
| SWE-Bench VerifiedAgentic coding | 72.8% | 77.2% | 72.0% | 73.1% |
| Tau-BenchAgentic tool use | 81.1% | Not in our dataset | Not in our dataset | Not in our dataset |
| Terminal-BenchAgentic terminal | 35.2% | Not in our dataset | Not in our dataset | 39.3% |
| ARC-AGI-2Visual reasoning | 9.9% | Not in our dataset | Not in our dataset | Not in our dataset |
Pricing & context Published USD rates per 1M tokens
| Specification | GPT-5.0 | 3.6 27B Thinking | 3.5 122B A10B Thinking | v3 3.2 Thinking |
|---|---|---|---|---|
| Input | $1.25 | $0.45 | $0.26 | $0.26 |
| Output | $10 | $2.7 | $2.08 | $0.38 |
| Context window | 400K | 262K | 262K | 164K |
| 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.