Model comparison
Claude 3 Haiku vs DeepSeek-V2 (MoE-236B, May 2024)
Claude 3 Haiku has enough public results to be ranked (#340); DeepSeek-V2 (MoE-236B, May 2024) does not yet, so treat this comparison as directional.
Last verified . 5 shared benchmarks.
DeepSeekSummary
- They share 5 benchmarks with published results for both. Claude 3 Haiku scores higher in 0 categories and DeepSeek-V2 (MoE-236B, May 2024) in 1 category; one gap is clear of the uncertainty.
- The widest gap is in coding, where DeepSeek-V2 (MoE-236B, May 2024) leads 40.4 to 26.4.
- The biggest single-benchmark swing is BigCodeBench Instruct: 39.4% for Claude 3 Haiku and 48.9% for DeepSeek-V2 (MoE-236B, May 2024).
- DeepSeek-V2 (MoE-236B, May 2024) has downloadable open weights; the other is API-only.
Side by side
| Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 25.9 | 40.3 |
| Released | 2024-03-07 | 2024-05-07 |
| Weights | Proprietary | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 37 | 10 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V2 (MoE-236B, May 2024) leads
Claude 3 Haiku: 26.4 (#325), DeepSeek-V2 (MoE-236B, May 2024): 40.4 (#139)
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| BigCodeBench Instruct | 39.4% | 48.9% |
| BigCodeBench Complete | 50.1% | 59.4% |
| WeirdML | 9.8% | — |
| LMArena Coding | 1199 | — |
| CadEval | 12% | — |
| HumanEval+ | 68.9% | — |
| MBPP+ | 68.8% | — |
Reasoning Not comparable
Claude 3 Haiku: 16.3 (#307), DeepSeek-V2 (MoE-236B, May 2024): —
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| Epoch Capabilities Index | 118.35 | 124.77 |
| WinoGrande | 74.2% | 86.3% |
| Kagi LLM Benchmark | 34.2% | — |
| LMArena Hard Prompts | 1174 | — |
| DTBench | 50.1% | — |
| LMCA | 8.8% | — |
| BIG-Bench Hard | — | 78.8% |
| ForecastBench | 53.2 | — |
| HellaSwag | — | 87.1% |
| PIQA | — | 83.9% |
Math Not comparable
Claude 3 Haiku: 9.8 (#319), DeepSeek-V2 (MoE-236B, May 2024): —
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.8% | — |
| LMArena Math | 1188 | — |
| MATH Level 5 | 14.9% | — |
Knowledge Not comparable
Claude 3 Haiku: 17.3 (#285), DeepSeek-V2 (MoE-236B, May 2024): —
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| MMLU | 73.8% | 78.4% |
| GPQA Diamond | 36.3% | — |
| Confabulations | 34.2% | — |
| LMArena Expert | 1148 | — |
| ARC (AI2) Challenge | — | 92.2% |
| TriviaQA | — | 80% |
Multimodal Not comparable
Claude 3 Haiku: 23.6 (#128), DeepSeek-V2 (MoE-236B, May 2024): —
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| LMArena Vision | 950 | — |
| ScienceQA | 72% | — |
Multilingual Not comparable
Claude 3 Haiku: 36.0 (#243), DeepSeek-V2 (MoE-236B, May 2024): —
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| LMArena Non-English | 1178 | — |
| LMArena Chinese | 1155 | — |
| LMArena French | 1195 | — |
| LMArena German | 1174 | — |
| LMArena Japanese | 1102 | — |
| LMArena Korean | 1109 | — |
| LMArena Russian | 1204 | — |
| LMArena Spanish | 1166 | — |
Instruction Following Not comparable
Claude 3 Haiku: 61.3 (#247), DeepSeek-V2 (MoE-236B, May 2024): —
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| LMArena Instruction Following | 1173 | — |
Long Context Not comparable
Claude 3 Haiku: 36.1 (#237), DeepSeek-V2 (MoE-236B, May 2024): —
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| LMArena Longer Query | 1190 | — |
Writing & Preference Not comparable
Claude 3 Haiku: 29.7 (#291), DeepSeek-V2 (MoE-236B, May 2024): —
| Benchmark | Claude 3 Haiku | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
| LMArena Text | 1195 | — |
| LMArena Creative Writing | 1157 | — |
| EQ-Bench Creative Writing | 717 | — |
| LMArena Multi-Turn | 1190 | — |
Frequently asked questions
Is Claude 3 Haiku better than DeepSeek-V2 (MoE-236B, May 2024)?
Claude 3 Haiku has enough public results to be ranked (#340); DeepSeek-V2 (MoE-236B, May 2024) does not yet, so treat this comparison as directional.
Is Claude 3 Haiku or DeepSeek-V2 (MoE-236B, May 2024) better for coding?
DeepSeek-V2 (MoE-236B, May 2024) scores higher on coding benchmarks: 40.4 versus 26.4 in the Noometry coding category.
How many benchmarks do Claude 3 Haiku and DeepSeek-V2 (MoE-236B, May 2024) share?
5 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and DeepSeek-V2 (MoE-236B, May 2024) has 10.