Model comparison
GPT-4 Turbo vs o1
o1 is the stronger model overall, scoring 40.9 to 30.5 on the Noometry Index. GPT-4 Turbo costs 1.8× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Last verified . 32 shared benchmarks.
Summary
- They share 32 benchmarks with published results for both. GPT-4 Turbo scores higher in 0 categories and o1 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where o1 leads 36.1 to 9.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.7% for GPT-4 Turbo and 73.3% for o1.
- GPT-4 Turbo is cheaper at $10 / $30 per million input/output tokens, against $15 / $60 for o1.
- o1 accepts more context: 200K tokens versus 128K.
Side by side
| GPT-4 Turbo | o1 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 30.5 | 40.9 |
| Released | 2023-11-06 | 2024-09-12 |
| Weights | Proprietary | Proprietary |
| Context window | 128K | 200K |
| Max output | 4K | 100K |
| Input $ / M tokens | $10 | $15 |
| Output $ / M tokens | $30 | $60 |
| Results tracked | 36 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1 leads
GPT-4 Turbo: 33.8 (#249), o1: 46.1 (#70)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| WeirdML | 18% | 47.6% |
| LMArena Coding | 1268 | 1367 |
| HumanEval+ | 86.6% | 89% |
| MBPP+ | 73.3% | 80.2% |
| Aider Polyglot | — | 61.7% |
| BigCodeBench Instruct | 48.2% | — |
| LiveBench Coding | — | 69.7% |
| BigCodeBench Complete | 58.2% | — |
| CadEval | — | 56% |
Agentic & Tool Use Not comparable
GPT-4 Turbo: —, o1: 24.6 (#117)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| METR Time Horizons | 36.7% | 51.1% |
| Cybench | — | 10% |
Reasoning o1 leads
GPT-4 Turbo: 15.3 (#317), o1: 27.9 (#111)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| SimpleBench | 25.1% | 41.7% |
| Chess Puzzles | 6% | 15% |
| LMArena Hard Prompts | 1251 | 1371 |
| DTBench | 61.6% | 74.7% |
| LMCA | 9.8% | 22.3% |
| Epoch Capabilities Index | 127.25 | 141.91 |
| ARC-AGI-1 | — | 30.7% |
| EnigmaEval | — | 5.7% |
| LiveBench Reasoning | — | 91.6% |
| LiveBench Data Analysis | — | 65.5% |
| ForecastBench | 59.4 | — |
| LiveBench | — | 75.7% |
Math o1 leads
GPT-4 Turbo: 9.0 (#322), o1: 36.1 (#175)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0.7% | 14.7% |
| OTIS Mock AIME 2024-2025 | 6.7% | 73.3% |
| LMArena Math | 1272 | 1388 |
| MATH Level 5 | 46.7% | 94.7% |
| LiveBench Math | — | 80.3% |
| FrontierMath (Feb 2025 set) | — | 9.3% |
Knowledge o1 leads
GPT-4 Turbo: 24.3 (#268), o1: 41.5 (#110)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| GPQA Diamond | 46.6% | 76.8% |
| Confabulations | 28.4% | 11.7% |
| LMArena Expert | 1223 | 1361 |
| Humanity's Last Exam | — | 8% |
| SimpleQA Verified | — | 41.1% |
| MMLU | 81.3% | — |
Multimodal o1 leads
GPT-4 Turbo: 30.6 (#110), o1: 34.2 (#93)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| LMArena Vision | 1090 | 1168 |
| GeoBench | — | 80% |
| VPCT | — | 37% |
| SpatialViz-Bench | — | 41.4% |
Multilingual o1 leads
GPT-4 Turbo: 40.5 (#216), o1: 48.6 (#142)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| LMArena Non-English | 1245 | 1358 |
| LMArena Chinese | 1242 | 1394 |
| LMArena French | 1276 | 1344 |
| LMArena German | 1259 | 1337 |
| LMArena Japanese | 1194 | 1346 |
| LMArena Korean | 1187 | 1396 |
| LMArena Russian | 1259 | 1356 |
| LMArena Spanish | 1260 | 1345 |
Instruction Following o1 leads
GPT-4 Turbo: 65.8 (#216), o1: 74.8 (#86)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| LMArena Instruction Following | 1249 | 1367 |
| LiveBench Instruction Following | — | 81.5% |
Long Context o1 leads
GPT-4 Turbo: 38.0 (#206), o1: 50.3 (#9)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| LMArena Longer Query | 1254 | 1378 |
| Fiction.LiveBench | — | 83.3% |
Writing & Preference o1 leads
GPT-4 Turbo: 47.7 (#206), o1: 55.6 (#144)
| Benchmark | GPT-4 Turbo | o1 |
|---|---|---|
| LMArena Text | 1272 | 1366 |
| LMArena Creative Writing | 1269 | 1348 |
| LMArena Multi-Turn | 1267 | 1369 |
| Short-Story Creative Writing | — | 70.2% |
| LiveBench Language | — | 65.4% |
Frequently asked questions
Is GPT-4 Turbo better than o1?
o1 is the stronger model overall, scoring 40.9 to 30.5 on the Noometry Index. GPT-4 Turbo costs 1.8× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Which is cheaper, GPT-4 Turbo or o1?
GPT-4 Turbo is cheaper. It lists at $10 per million input tokens and $30 per million output tokens; o1 lists at $15 and $60.
Is GPT-4 Turbo or o1 better for coding?
o1 scores higher on coding benchmarks: 46.1 versus 33.8 in the Noometry coding category.
Which has the bigger context window?
o1 does, with 200K tokens against 128K.
How many benchmarks do GPT-4 Turbo and o1 share?
32 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and o1 has 52.