Model comparison
DeepSeek-R1 vs GPT-5.1
GPT-5.1 is the stronger model overall, scoring 49.0 to 42.3 on the Noometry Index. DeepSeek-R1 costs 3.8× less per token, which makes it the better buy when GPT-5.1's lead doesn't matter for your workload.
Last verified . 42 shared benchmarks.
DeepSeekSummary
- They share 42 benchmarks with published results for both. DeepSeek-R1 scores higher in 0 categories and GPT-5.1 in 9 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.1 leads 39.8 to 18.6.
- The biggest single-benchmark swing is ARC-AGI-1: 21.2% for DeepSeek-R1 and 72.8% for GPT-5.1.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $1.25 / $10 for GPT-5.1.
- GPT-5.1 accepts more context: 400K tokens versus 164K.
Side by side
| DeepSeek-R1 | GPT-5.1 | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.3 | 49.0 |
| Released | 2025-01-20 | 2025-11-13 |
| Weights | Proprietary | Proprietary |
| Context window | 164K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $0.50 | $1.25 |
| Output $ / M tokens | $2.15 | $10 |
| Results tracked | 52 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
DeepSeek-R1: 46.3 (#68), GPT-5.1: 46.4 (#66)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| SciCode | 35.7% | 43.3% |
| WeirdML | 41.6% | 60.8% |
| LiveBench Coding | 66.7% | 72.5% |
| LMArena Coding | 1427 | 1454 |
| ALE-Bench | 804.12 | 1,192 |
| SWE-bench Verified | — | 68% |
| SWE-bench Verified (bash only) | — | 66% |
| Aider Polyglot | 71.4% | — |
| LMArena WebDev | — | 1395 |
| GSO | — | 13.7% |
| AlgoTune | 1.7 | — |
Agentic & Tool Use GPT-5.1 leads
DeepSeek-R1: 30.7 (#75), GPT-5.1: 32.7 (#60)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| DeepResearch Bench | 35.1% | 42.8% |
| Terminal-Bench | — | 47.6% |
| BALROG | 34.9% | — |
| LMArena Search | — | 1199 |
| METR Time Horizons | 53.8% | — |
| Vending-Bench 2 | — | 1,473 |
Reasoning GPT-5.1 leads
DeepSeek-R1: 18.6 (#278), GPT-5.1: 39.8 (#58)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| ARC-AGI-2 | 1.3% | 17.6% |
| SimpleBench | 40.8% | 53.2% |
| ARC-AGI-1 | 21.2% | 72.8% |
| CritPt | 1.1% | 4.9% |
| LiveBench Reasoning | 83.2% | 95.8% |
| LMArena Hard Prompts | 1416 | 1457 |
| LiveBench Data Analysis | 69.8% | 72.1% |
| Epoch Capabilities Index | 141.29 | 149.64 |
| ForecastBench | 60 | 58.1 |
| LiveBench | 71.6% | 78.8% |
| Kagi LLM Benchmark | 69.4% | — |
| Chess Puzzles | — | 32% |
| EnigmaEval | — | 11.2% |
| Mystery Game Puzzles | — | 19% |
| DTBench | — | 90.1% |
| LMCA | — | 43.9% |
Math GPT-5.1 leads
DeepSeek-R1: 43.8 (#79), GPT-5.1: 52.2 (#51)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 88.6% |
| Omni-MATH | 42.4% | 46.4% |
| LiveBench Math | 80.7% | 94.5% |
| LMArena Math | 1400 | 1447 |
| MATH Level 5 | 96.6% | — |
| FrontierMath (Feb 2025 set) | — | 31% |
| FrontierMath Tier 4 (v1) | — | 12.5% |
Knowledge GPT-5.1 leads
DeepSeek-R1: 44.5 (#87), GPT-5.1: 50.6 (#71)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| GPQA Diamond | 76.3% | 87.6% |
| MMLU-Pro | 79.3% | 57.9% |
| Vectara Hallucination Rate | 11.3% | 10.9% |
| GPQA (HELM) | 66.6% | 44.2% |
| LMArena Expert | 1394 | 1470 |
| Humanity's Last Exam | — | 23.7% |
| SimpleQA Verified | — | 48% |
| Confabulations | 12.7% | — |
Multimodal Not comparable
DeepSeek-R1: —, GPT-5.1: 44.8 (#19)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| LMArena Vision | — | 1250 |
| VPCT | — | 58.7% |
| LMArena Document | — | 1403 |
Multilingual GPT-5.1 leads
DeepSeek-R1: 52.4 (#85), GPT-5.1: 53.8 (#56)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| LMArena Non-English | 1412 | 1431 |
| LMArena Chinese | 1442 | 1495 |
| LMArena French | 1417 | 1450 |
| LMArena German | 1404 | 1438 |
| LMArena Japanese | 1391 | 1453 |
| LMArena Korean | 1360 | 1401 |
| LMArena Russian | 1423 | 1435 |
| LMArena Spanish | 1411 | 1433 |
Instruction Following GPT-5.1 leads
DeepSeek-R1: 72.0 (#143), GPT-5.1: 83.9 (#1)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| LiveBench Instruction Following | 80.5% | 93.3% |
| IFEval | 78.4% | 93.5% |
| LMArena Instruction Following | 1382 | 1443 |
Long Context GPT-5.1 leads
DeepSeek-R1: 45.4 (#36), GPT-5.1: 47.6 (#14)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| LMArena Longer Query | 1391 | 1447 |
| Fiction.LiveBench | 75% | — |
| CL-bench | — | 23.7% |
| CL-bench Life | — | 17.3% |
Writing & Preference GPT-5.1 leads
DeepSeek-R1: 61.4 (#88), GPT-5.1: 64.5 (#55)
| Benchmark | DeepSeek-R1 | GPT-5.1 |
|---|---|---|
| LMArena Text | 1428 | 1443 |
| LMArena Creative Writing | 1405 | 1427 |
| WildBench | 82.8% | 86.3% |
| LMArena Multi-Turn | 1405 | 1450 |
| LiveBench Language | 48.5% | 80.2% |
| Short-Story Creative Writing | 83% | — |
| EQ-Bench Creative Writing | 1500 | — |
Frequently asked questions
Is DeepSeek-R1 better than GPT-5.1?
GPT-5.1 is the stronger model overall, scoring 49.0 to 42.3 on the Noometry Index. DeepSeek-R1 costs 3.8× less per token, which makes it the better buy when GPT-5.1's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-R1 or GPT-5.1?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; GPT-5.1 lists at $1.25 and $10.
Is DeepSeek-R1 or GPT-5.1 better for coding?
They score almost the same on coding (46.3 vs 46.4); test both on your own repository before choosing.
Which has the bigger context window?
GPT-5.1 does, with 400K tokens against 164K.
How many benchmarks do DeepSeek-R1 and GPT-5.1 share?
42 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and GPT-5.1 has 63.