cheapaimodelSearch

DeepSeek-V4.1-Flash vs Kimi K3

deepseek-v4-1-flashkimi-k3
API docsGet API key

DeepSeek-V4.1-Flash costs $0.209351 per 1M input tokens and $0.837404 per 1M output tokens, with a 1M-token context window; Kimi K3 costs $2.093511 per 1M input tokens and $10.467555 per 1M output tokens, with a 1M-token context window.

Side by side

Prices in US dollars; limits and capabilities as published by each model’s provider.

DeepSeek-V4.1-Flash and Kimi K3 side by side
SpecDeepSeek-V4.1-FlashKimi K3
Input/ 1M tokens$0.209351$2.093511
Output/ 1M tokens$0.837404$10.467555
Cache read/ 1M tokens$0.004187$0.209351
Cache write/ 1M tokens$0$2.093511
Discount30% offList price $0.30 / $1.2030% offList price $3 / $15
Contexttokens1M1M
Max outputtokens384K1M
Input typestext, imagetext, image, video
Output typestexttext
Released2026-09-102026-07-16
Tokens, last 7 daystokens82M—
Endpoints/v1/chat/completions/v1/responses/v1/chat/completions/v1/responses

Same workload

One workload of 1,000,000 input and 200,000 output tokens, priced at each model’s current rates.

DeepSeek-V4.1-Flash
$0.3768$0.2094 input + $0.1675 output
Kimi K3
$4.187$2.0935 input + $2.0935 output

For this workload, DeepSeek-V4.1-Flash costs $3.8102 less than Kimi K3.

Same code

Only the model field changes.

curl https://api.cheapaimodel.ai/v1/chat/completions \
  -H "Authorization: Bearer $CHEAPAIMODEL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-1-flash",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

When to choose which

Differences in the published specs, then each provider’s own description.

  • DeepSeek-V4.1-Flash has the lower input price ($0.209351 vs $2.093511 per 1M tokens).
  • DeepSeek-V4.1-Flash has the lower output price ($0.837404 vs $10.467555 per 1M tokens).
  • Kimi K3 accepts video input; DeepSeek-V4.1-Flash doesn’t.
  • DeepSeek-V4.1-Flash was released more recently (2026-09-10 vs 2026-07-16).
  • Kimi K3 can return more output in one response (1M vs 384K tokens).

About DeepSeek-V4.1-Flash

The smallest model in DeepSeek's new architecture family, with native image understanding. It runs in thinking mode by default and also offers a non-thinking mode; DeepSeek reports it outperforms its earlier DeepSeek-V4-Pro flagship on benchmarks.

About Kimi K3

Moonshot AI's flagship open-weight model, with 2.8 trillion parameters built on Kimi Delta Attention and Attention Residuals. It has native visual understanding and is designed for long-horizon coding, knowledge work and deep reasoning.

API docsGet API key