Kimi K2 Instruct 0905
Provider: deepinfra ·
deepinfra/moonshotai/Kimi-K2-Instruct-0905 Pricing per million tokens
| Component | USD per 1M tokens |
|---|---|
| Input | $0.50 |
| Output | $2.00 |
| Cached input (read) | $0.40 |
💡 With prompt caching, you save up to 20% on cached input tokens — massive for repeated context like system prompts, RAG retrieval, or long conversations.
Monthly cost estimates
Assuming 30% prompt-cache hit rate where available. Adjust for your actual usage.
Small
1M in / 200k out
$0.87
per month
Saves $0.03 via caching
Medium
10M in / 2000k out
$8.70
per month
Saves $0.30 via caching
Large
100M in / 20000k out
$87.00
per month
Saves $3.00 via caching
Context
- Input window
- 262,144 tokens
- Max output
- 262,144 tokens
Capabilities
- ⬜ Vision (image input)
- ✅ Function / tool calling
- ⬜ Prompt caching
- ⬜ Web search
- ⬜ JSON / response schema
Compare Kimi K2 Instruct 0905 with similar models
Hermes 3 Llama 3.1 405B
deepinfra
$1.00 in
/ $1.00 out
Hermes 3 Llama 3.1 70B
deepinfra
$0.30 in
/ $0.30 out
Qwen3 235B A22B Thinking 2507
deepinfra
$0.30 in
/ $2.90 out
Qwen3 Coder 480B A35B Instruct
deepinfra
$0.40 in
/ $1.60 out
Qwen3 Coder 480B A35B Instruct Turbo
deepinfra
$0.29 in
/ $1.20 out
L3.1 70B Euryale V2.2
deepinfra
$0.65 in
/ $0.75 out
Pricing data sourced from LiteLLM and refreshed regularly. Last updated May 20, 2026. Always verify with the provider's official pricing page before making business decisions.