Model comparison
Claude 3.5 Haiku vs DeepSeek-R1
DeepSeek-R1 is the stronger model overall, scoring 42.0 to 28.9 on the Noometry Index.
Last verified . 41 shared benchmarks.
DeepSeekSummary
- They share 41 benchmarks with published results for both. Claude 3.5 Haiku scores higher in 0 categories and DeepSeek-R1 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-R1 leads 42.0 to 13.7.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 4.3% for Claude 3.5 Haiku and 66.4% for DeepSeek-R1.
Side by side
| Claude 3.5 Haiku | DeepSeek-R1 | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 28.9 | 42.0 |
| Released | 2024-10-22 | 2025-01-20 |
| Weights | Proprietary | Proprietary |
| Context window | — | 164K |
| Max output | — | 64K |
| Input $ / M tokens | — | $0.50 |
| Output $ / M tokens | — | $2.15 |
| Results tracked | 49 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
Claude 3.5 Haiku: 32.6 (#258), DeepSeek-R1: 45.9 (#63)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| Aider Polyglot | 28% | 71.4% |
| SciCode | 27.4% | 35.7% |
| WeirdML | 30.7% | 41.6% |
| LiveBench Coding | 51.4% | 66.7% |
| LMArena Coding | 1287 | 1427 |
| BigCodeBench Instruct | 46.1% | — |
| BigCodeBench Complete | 59% | — |
| CadEval | 32% | — |
| ALE-Bench | — | 804.12 |
| AlgoTune | — | 1.7 |
Agentic & Tool Use DeepSeek-R1 leads
Claude 3.5 Haiku: 27.7 (#90), DeepSeek-R1: 30.4 (#68)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| BALROG | 19.3% | 34.9% |
| DeepResearch Bench | — | 35.1% |
| METR Time Horizons | — | 53.8% |
Reasoning DeepSeek-R1 leads
Claude 3.5 Haiku: 17.1 (#286), DeepSeek-R1: 18.7 (#253)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| CritPt | 0% | 1.1% |
| LiveBench Reasoning | 28.1% | 83.2% |
| LMArena Hard Prompts | 1251 | 1415 |
| LiveBench Data Analysis | 48.5% | 69.8% |
| Epoch Capabilities Index | 127.15 | 141.3 |
| LiveBench | 43.5% | 71.6% |
| ARC-AGI-2 | — | 1.3% |
| SimpleBench | — | 40.8% |
| Kagi LLM Benchmark | — | 69.4% |
| ARC-AGI-1 | — | 21.2% |
| DTBench | 56.7% | — |
| ForecastBench | — | 60 |
Math DeepSeek-R1 leads
Claude 3.5 Haiku: 13.7 (#294), DeepSeek-R1: 42.0 (#70)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 4.3% | 66.4% |
| Omni-MATH | 22.4% | 42.4% |
| LiveBench Math | 35.5% | 80.7% |
| LMArena Math | 1244 | 1401 |
| MATH Level 5 | 46.4% | 96.6% |
| FrontierMath (Feb 2025 set) | 0.3% | — |
Knowledge DeepSeek-R1 leads
Claude 3.5 Haiku: 18.7 (#276), DeepSeek-R1: 44.3 (#81)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| GPQA Diamond | 38.1% | 76.3% |
| MMLU-Pro | 60.5% | 79.3% |
| Confabulations | 36.7% | 12.7% |
| GPQA (HELM) | 36.3% | 66.6% |
| LMArena Expert | 1208 | 1394 |
| Vectara Hallucination Rate | — | 11.3% |
| MMLU | 74.3% | — |
Multimodal Not comparable
Claude 3.5 Haiku: 26.3 (#113), DeepSeek-R1: —
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| LMArena Vision | 1092 | — |
| GeoBench | 34% | — |
Multilingual DeepSeek-R1 leads
Claude 3.5 Haiku: 40.2 (#213), DeepSeek-R1: 52.6 (#80)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| LMArena Non-English | 1238 | 1412 |
| LMArena Chinese | 1229 | 1443 |
| LMArena French | 1262 | 1415 |
| LMArena German | 1237 | 1405 |
| LMArena Japanese | 1175 | 1394 |
| LMArena Korean | 1172 | 1359 |
| LMArena Russian | 1253 | 1423 |
| LMArena Spanish | 1261 | 1410 |
Instruction Following DeepSeek-R1 leads
Claude 3.5 Haiku: 63.1 (#229), DeepSeek-R1: 72.2 (#138)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| LiveBench Instruction Following | 61.9% | 80.5% |
| IFEval | 79.2% | 78.4% |
| LMArena Instruction Following | 1241 | 1382 |
Long Context DeepSeek-R1 leads
Claude 3.5 Haiku: 38.3 (#195), DeepSeek-R1: 45.4 (#36)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| LMArena Longer Query | 1261 | 1391 |
| Fiction.LiveBench | — | 75% |
Writing & Preference DeepSeek-R1 leads
Claude 3.5 Haiku: 42.9 (#228), DeepSeek-R1: 61.5 (#84)
| Benchmark | Claude 3.5 Haiku | DeepSeek-R1 |
|---|---|---|
| LMArena Text | 1255 | 1428 |
| LMArena Creative Writing | 1233 | 1406 |
| Short-Story Creative Writing | 73.5% | 83% |
| EQ-Bench Creative Writing | 1146 | 1500 |
| WildBench | 76% | 82.8% |
| LMArena Multi-Turn | 1265 | 1404 |
| LiveBench Language | 35.4% | 48.5% |
Frequently asked questions
Is Claude 3.5 Haiku better than DeepSeek-R1?
DeepSeek-R1 is the stronger model overall, scoring 42.0 to 28.9 on the Noometry Index.
Is Claude 3.5 Haiku or DeepSeek-R1 better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 45.9 versus 32.6 in the Noometry coding category.
How many benchmarks do Claude 3.5 Haiku and DeepSeek-R1 share?
41 benchmarks have published results for both models. Claude 3.5 Haiku has 49 scored results on Noometry and DeepSeek-R1 has 52.