Compare · ModelsLive · 2 picked · head to head
DeepSeek-V2 (MoE-236B, May 2024) vs PaLM 2-L
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
DeepSeek-V2 (MoE-236B, May 2024) wins on 3/4 benchmarks
DeepSeek-V2 (MoE-236B, May 2024) wins 3 of 4 shared benchmarks. Leads in knowledge.
Category leads
knowledge·DeepSeek-V2 (MoE-236B, May 2024)
Hype vs Reality
Attention vs performance
DeepSeek-V2 (MoE-236B, May 2024)
#9 by perf·no signal
PaLM 2-L
#13 by perf·no signal
Best value
Pricing unknown
DeepSeek-V2 (MoE-236B, May 2024)
n/a
no price
PaLM 2-L
n/a
no price
Vendor risk
Mixed exposure
One or more vendors flagged
DeepSeek
$3.4B·Tier 1
U
Unknown
private · undisclosed
Head to head
4 benchmarks · 2 models
DeepSeek-V2 (MoE-236B, May 2024)PaLM 2-L
ARC AI2
DeepSeek-V2 (MoE-236B, May 2024) leads by +30.7
AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval.
DeepSeek-V2 (MoE-236B, May 2024)
89.6
PaLM 2-L
58.9
HellaSwag
DeepSeek-V2 (MoE-236B, May 2024) leads by +0.4
HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios.
DeepSeek-V2 (MoE-236B, May 2024)
82.8
PaLM 2-L
82.4
TriviaQA
PaLM 2-L leads by +6.1
TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents.
DeepSeek-V2 (MoE-236B, May 2024)
80.0
PaLM 2-L
86.1
Winogrande
DeepSeek-V2 (MoE-236B, May 2024) leads by +6.6
WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs.
DeepSeek-V2 (MoE-236B, May 2024)
72.6
PaLM 2-L
66.0
Full benchmark table
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | PaLM 2-L |
|---|---|---|
ARC AI2 AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval. | 89.6 | 58.9 |
HellaSwag HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios. | 82.8 | 82.4 |
TriviaQA TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents. | 80.0 | 86.1 |
Winogrande WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs. | 72.6 | 66.0 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| — | — | — | — | |
U PaLM 2-L | — | — | — | — |
People also compared