Kimi and more
Moonshot AI
from $0.875 / 1M
9 models · typical $1.34 · top 93.1% GPQA, max reasoning
Developer head to head · prices updated Nov 1, 2026
Moonshot AI vs DeepSeek
Kimi vs DeepSeek: every model's API price, compared
DeepSeek starts cheaper ($0.025 vs $0.875 per 1M tokens); DeepSeek's typical model costs less ($0.30 vs $1.34); Moonshot AI has the more accurate top model (93.1% vs 91.7% GPQA).
DeepSeek and more
DeepSeek
from $0.025 / 1M
23 models · typical $0.30 · top 91.7% GPQA, max reasoning
- Cheaper entry modelDeepSeek
- More accurate top modelMoonshot AI
- Cheaper 85%+ GPQA modelDeepSeek
- More modelsDeepSeek
The matchups that matter
| Moonshot AI | DeepSeek | |
|---|---|---|
| Cheapest model | Kimi K2 Instruct $0.875 via DeepInfra | deepseek-r1-distill-llama-8b $0.025 via Nscale |
| Most accurate (GPQA) | Kimi K3 $5.40 via SiliconFlow · 93.1% GPQA, max reasoning | DeepSeek V4 Pro 0813 $1.63 via DeepInfra · 91.7% GPQA, max reasoning |
| Cheapest with 85%+ GPQA | Kimi K2.6 $1.34 via W&B Inference (CoreWeave) · 90.8% GPQA | DeepSeek V4 Flash 0731 $0.105 via DeepInfra · 91.0% GPQA, max reasoning |
| Typical model price | $1.34 | $0.30 |
| Models we track | 9 | 23 |
Flagships: Kimi K3 vs DeepSeek V4 Pro 0813Cheapest: Kimi K2 Instruct vs deepseek-r1-distill-llama-8bKimi K3 vs DeepSeek V4 Pro
Blended list price per 1M tokens (3 input : 1 output) at each model's cheapest provider; only models at least two providers sell. GPQA Diamond by Epoch AI (CC BY 4.0), best measured reasoning setting.
Every model, cheapest first
Moonshot AI
- Kimi K2 Instruct$0.875 via DeepInfra
- kimi-k2-instruct-0905$0.875 via DeepInfra
- Kimi K2.5$0.90 via DeepInfra
- Kimi K2 Thinking$1.08 via Fireworks AI
- Kimi K2.690.8% GPQA$1.34 via W&B Inference (CoreWeave)
- Kimi K2.7 Code87.9% GPQA$1.36 via DeepInfra
- Kimi K3.1 Pro$4.00 via Moonshot AI
- Kimi K393.1% GPQA, max reasoning$5.40 via SiliconFlow
- Kimi K3.1 Pro Highspeed$8.00 via Moonshot AI
DeepSeek
- deepseek-r1-distill-llama-8b$0.025 via Nscale
- deepseek-r1-distill-qwen-14b44.7% GPQA$0.07 via Nscale
- deepseek-r1-distill-qwen-1.5b33.6% GPQA$0.09 via Nscale
- DeepSeek V4 Flash 073191.0% GPQA, max reasoning$0.105 via DeepInfra
- DeepSeek-V4-Flash$0.1125 via DeepInfra
- Deepseek R1 Distill Qwen 32B64.1% GPQA$0.15 via Nscale
- deepseek-r1-distill-qwen-7b$0.20 via Fireworks AI
- DeepSeek V4.1 Flash 1017$0.21 via DeepInfra
- DeepSeek-V4.1-Flash89.8% GPQA, max reasoning$0.21 via DeepInfra
- DeepSeek-V3.2$0.29 via DeepInfra
- DeepSeek-R1-052876.3% GPQA$0.30 via Lambda
- deepseek-r1-distill-llama-70b55.7% GPQA$0.30 via DeepInfra
Questions people ask
Is Moonshot AI cheaper than DeepSeek?
No at the cheap end: DeepSeek's cheapest model, deepseek-r1-distill-llama-8b, starts at $0.025 per 1M tokens, against $0.875 for Moonshot AI's Kimi K2 Instruct. The typical (median) model costs $1.34 at Moonshot AI and $0.30 at DeepSeek.
Which has the more accurate model, Moonshot AI or DeepSeek?
Moonshot AI's most accurate model we track, Kimi K3, scores 93.1% GPQA, max reasoning (GPQA Diamond, Epoch AI) for $5.40 per 1M tokens; DeepSeek's DeepSeek V4 Pro 0813 scores 91.7% GPQA, max reasoning for $1.63.
How many Moonshot AI and DeepSeek models can I buy through APIs?
We track 9 Moonshot AI and 23 DeepSeek models that at least two providers sell, with prices from each provider's pricing page.