Noometry

Model comparison

Claude 3 Haiku vs Deepseek Coder v2

Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 25.9 on the Noometry Index.

Last verified . 22 shared benchmarks.

Claude 3 Haiku Anthropic

25.9

Rank #340 Confirmed

Deepseek Coder v2 DeepSeek

35.9

Rank #220 Confirmed

Summary

  • They share 22 benchmarks with published results for both. Claude 3 Haiku scores higher in 0 categories and Deepseek Coder v2 in 8 categories; 5 gaps are clear of the uncertainty.
  • The widest gap is in math, where Deepseek Coder v2 leads 34.9 to 9.8.
  • The biggest single-benchmark swing is BigCodeBench Complete: 50.1% for Claude 3 Haiku and 59.7% for Deepseek Coder v2.
  • Deepseek Coder v2 has downloadable open weights; the other is API-only.

Side by side

Claude 3 HaikuDeepseek Coder v2
ProviderAnthropicDeepSeek
Noometry Index25.935.9
Released2024-03-072024-06-17
WeightsProprietaryOpen
Context window——
Max output——
Input $ / M tokens——
Output $ / M tokens——
Results tracked3724

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Deepseek Coder v2 leads

Claude 3 Haiku: 26.4 (#325), Deepseek Coder v2: 38.1 (#183)

BenchmarkClaude 3 HaikuDeepseek Coder v2
BigCodeBench Instruct39.4%48.2%
LMArena Coding11991251
BigCodeBench Complete50.1%59.7%
HumanEval+68.9%82.3%
MBPP+68.8%75.1%
WeirdML9.8%—
CadEval12%—

Reasoning Deepseek Coder v2 leads

Claude 3 Haiku: 16.3 (#307), Deepseek Coder v2: 23.6 (#176)

BenchmarkClaude 3 HaikuDeepseek Coder v2
LMArena Hard Prompts11741207
WinoGrande74.2%83.7%
Kagi LLM Benchmark34.2%—
DTBench50.1%—
LMCA8.8%—
Epoch Capabilities Index118.35—
ForecastBench53.2—

Math Deepseek Coder v2 leads

Claude 3 Haiku: 9.8 (#319), Deepseek Coder v2: 34.9 (#190)

BenchmarkClaude 3 HaikuDeepseek Coder v2
LMArena Math11881241
OTIS Mock AIME 2024-20251.8%—
MATH Level 514.9%—
GSM8K—94.5%

Knowledge Deepseek Coder v2 leads

Claude 3 Haiku: 17.3 (#285), Deepseek Coder v2: 32.3 (#212)

BenchmarkClaude 3 HaikuDeepseek Coder v2
LMArena Expert11481181
GPQA Diamond36.3%—
Confabulations34.2%—
ARC (AI2) Challenge—64.3%
MMLU73.8%—

Multimodal Not comparable

Claude 3 Haiku: 23.6 (#128), Deepseek Coder v2: —

BenchmarkClaude 3 HaikuDeepseek Coder v2
LMArena Vision950—
ScienceQA72%—

Multilingual Too close to call

Claude 3 Haiku: 36.0 (#243), Deepseek Coder v2: 36.3 (#240)

BenchmarkClaude 3 HaikuDeepseek Coder v2
LMArena Non-English11781182
LMArena Chinese11551201
LMArena French11951185
LMArena German11741164
LMArena Japanese11021126
LMArena Korean11091104
LMArena Russian12041188
LMArena Spanish11661153

Instruction Following Too close to call

Claude 3 Haiku: 61.3 (#247), Deepseek Coder v2: 61.7 (#242)

BenchmarkClaude 3 HaikuDeepseek Coder v2
LMArena Instruction Following11731180

Long Context Too close to call

Claude 3 Haiku: 36.1 (#237), Deepseek Coder v2: 37.0 (#224)

BenchmarkClaude 3 HaikuDeepseek Coder v2
LMArena Longer Query11901219

Writing & Preference Deepseek Coder v2 leads

Claude 3 Haiku: 29.7 (#291), Deepseek Coder v2: 38.2 (#253)

BenchmarkClaude 3 HaikuDeepseek Coder v2
LMArena Text11951191
LMArena Creative Writing11571120
LMArena Multi-Turn11901177
EQ-Bench Creative Writing717—

Frequently asked questions

Is Claude 3 Haiku better than Deepseek Coder v2?

Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 25.9 on the Noometry Index.

Is Claude 3 Haiku or Deepseek Coder v2 better for coding?

Deepseek Coder v2 scores higher on coding benchmarks: 38.1 versus 26.4 in the Noometry coding category.

How many benchmarks do Claude 3 Haiku and Deepseek Coder v2 share?

22 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and Deepseek Coder v2 has 24.

Related comparisons

Go deeper