Price Per TokenPrice Per Token
Deepseek
Deepseek
vs
Moonshotai
Moonshotai

DeepSeek V3 vs Kimi K2 Thinking

A detailed comparison of pricing, benchmarks, and capabilities

OpenClaw

Best LLMs for OpenClaw Vote for which model works best with OpenClaw

112 out of our 301 tracked models have had a price change in February.

Get our weekly newsletter on pricing changes, new releases, and tools.

Key Takeaways

DeepSeek V3 wins:

  • Cheaper input tokens
  • Cheaper output tokens

Kimi K2 Thinking wins:

  • Larger context window
  • Faster response time
  • Higher intelligence benchmark
  • Better at coding
  • Better at math
  • Has reasoning mode
Price Advantage
DeepSeek V3
Benchmark Advantage
Kimi K2 Thinking
Context Window
Kimi K2 Thinking
Speed
Kimi K2 Thinking

Pricing Comparison

Price Comparison

MetricDeepSeek V3Kimi K2 ThinkingWinner
Input (per 1M tokens)$0.30$0.40 DeepSeek V3
Output (per 1M tokens)$1.20$1.75 DeepSeek V3
Cache Read (per 1M)$150000.00$200000.00 DeepSeek V3
Using a 3:1 input/output ratio, DeepSeek V3 is 29% cheaper overall.

DeepSeek V3 Providers

DeepSeek $0.01 (Cheapest)
Chutes $0.30
DeepInfra $0.32
Novita $0.40

Kimi K2 Thinking Providers

Chutes $0.40 (Cheapest)
DeepInfra $0.47
Parasail $0.50
SiliconFlow $0.55
AtlasCloud $0.60

Benchmark Comparison

7
Benchmarks Compared
0
DeepSeek V3 Wins
0
Kimi K2 Thinking Wins

Benchmark Scores

BenchmarkDeepSeek V3Kimi K2 ThinkingWinner
Intelligence Index
Overall intelligence score
-40.7-
Coding Index
Code generation & understanding
-34.8-
Math Index
Mathematical reasoning
-94.7-
MMLU Pro
Academic knowledge
-84.8-
GPQA
Graduate-level science
-83.8-
LiveCodeBench
Competitive programming
-85.3-
Aider
Real-world code editing
48.4--
Kimi K2 Thinking significantly outperforms in coding benchmarks.

Cost vs Quality

X-axis:
Y-axis:
Loading chart...
Other models

Context & Performance

Context Window

DeepSeek V3
163,840
tokens
Max output: 163,840 tokens
Kimi K2 Thinking
262,144
tokens
Max output: 65,535 tokens
Kimi K2 Thinking has a 38% larger context window.

Speed Performance

MetricDeepSeek V3Kimi K2 ThinkingWinner
Tokens/secondN/A82.1 tok/s
Time to First TokenN/A0.64s

Capabilities

Feature Comparison

FeatureDeepSeek V3Kimi K2 Thinking
Vision (Image Input)
Tool/Function Calls
Reasoning Mode
Audio Input
Audio Output
PDF Input
Prompt Caching
Web Search

License & Release

PropertyDeepSeek V3Kimi K2 Thinking
LicenseOpen SourceProprietary
AuthorDeepseekMoonshotai
ReleasedDec 2024Nov 2025

DeepSeek V3 Modalities

Input
text
Output
text

Kimi K2 Thinking Modalities

Input
text
Output
text

Related Comparisons

Compare DeepSeek V3 with:

Compare Kimi K2 Thinking with:

Frequently Asked Questions

DeepSeek V3 has cheaper input pricing at $0.30/M tokens. DeepSeek V3 has cheaper output pricing at $1.20/M tokens.
Kimi K2 Thinking scores higher on coding benchmarks with a score of 34.8, compared to DeepSeek V3's score of N/A.
DeepSeek V3 has a 163,840 token context window, while Kimi K2 Thinking has a 262,144 token context window.
DeepSeek V3 does not support vision. Kimi K2 Thinking does not support vision.