Compare · ModelsLive · 2 picked · head to head
DeepSeek V3 vs DeepSeek-V2 (MoE-236B, May 2024)
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
DeepSeek V3 wins on 6/7 benchmarks
DeepSeek V3 wins 6 of 7 shared benchmarks. Leads in knowledge · reasoning.
Category leads
knowledge·DeepSeek V3reasoning·DeepSeek V3
Hype vs Reality
Attention vs performance
DeepSeek V3
#43 by perf·no signal
DeepSeek-V2 (MoE-236B, May 2024)
#8 by perf·no signal
Best value
DeepSeek V3
DeepSeek V3
97.5 pts/$
$0.60/M
DeepSeek-V2 (MoE-236B, May 2024)
—
no price
Vendor risk
Mixed exposure
One or more vendors flagged
DeepSeek
$3.4B·Tier 1
DeepSeek
$3.4B·Tier 1
Head to head
7 benchmarks · 2 models
DeepSeek V3DeepSeek-V2 (MoE-236B, May 2024)
ARC AI2
DeepSeek V3 leads by +4.1
AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval.
DeepSeek V3
93.7
DeepSeek-V2 (MoE-236B, May 2024)
89.6
BBH
DeepSeek V3 leads by +11.6
BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans.
DeepSeek V3
83.3
DeepSeek-V2 (MoE-236B, May 2024)
71.7
HellaSwag
DeepSeek V3 leads by +2.4
HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios.
DeepSeek V3
85.2
DeepSeek-V2 (MoE-236B, May 2024)
82.8
MMLU
DeepSeek V3 leads by +11.7
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
DeepSeek V3
82.9
DeepSeek-V2 (MoE-236B, May 2024)
71.2
PIQA
DeepSeek V3 leads by +1.6
PIQA (Physical Interaction QA) · tests intuitive physical reasoning by asking models to select the correct approach for everyday physical tasks.
DeepSeek V3
69.4
DeepSeek-V2 (MoE-236B, May 2024)
67.8
TriviaQA
DeepSeek V3 leads by +2.9
TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents.
DeepSeek V3
82.9
DeepSeek-V2 (MoE-236B, May 2024)
80.0
Winogrande
DeepSeek-V2 (MoE-236B, May 2024) leads by +2.2
WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs.
DeepSeek V3
70.4
DeepSeek-V2 (MoE-236B, May 2024)
72.6
Full benchmark table
| Benchmark | DeepSeek V3 | DeepSeek-V2 (MoE-236B, May 2024) |
|---|---|---|
ARC AI2 AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval. | 93.7 | 89.6 |
BBH BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans. | 83.3 | 71.7 |
HellaSwag HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios. | 85.2 | 82.8 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 82.9 | 71.2 |
PIQA PIQA (Physical Interaction QA) · tests intuitive physical reasoning by asking models to select the correct approach for everyday physical tasks. | 69.4 | 67.8 |
TriviaQA TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents. | 82.9 | 80.0 |
Winogrande WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs. | 70.4 | 72.6 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| DeepSeek V3 | $0.32 | $0.89 | 164K tokens (~82 books) | $4.63 |
| DeepSeek-V2 (MoE-236B, May 2024) | — | — | — | — |
People also compared
DeepSeek V3 vs GPT-4oDeepSeek V3 vs Qwen2.5 Coder 32B InstructDeepSeek-V2 (MoE-236B, May 2024) vs GPT-5 ChatClaude Mythos Preview vs DeepSeek-V2 (MoE-236B, May 2024)DeepSeek-V2 (MoE-236B, May 2024) vs Qwen3.5 397B A17BDeepSeek-V2 (MoE-236B, May 2024) vs DeepSeek V3.2 SpecialeClaude Instant vs DeepSeek-V2 (MoE-236B, May 2024)DeepSeek-V2 (MoE-236B, May 2024) vs Step 3.5 Flash