Decision in 20 seconds
As of July 2, 2026, Kimi/Moonshot model selection starts with the model and pricing table: K2.7 Code, K2.7 Code HighSpeed, K2.6, K2.5, Moonshot V1, Kimi Code, and DeepSeek-V4-Flash as the outside API cost baseline.
Use this page when
- You need model names, prices, context windows, and Kimi Code workflow notes.
This page is not for
- Performance claims without official or reproducible evidence.
- Enterprise purchasing decisions that require private contract terms.
Key points
- `kimi-k2.7-code`: $0.19 cache-hit input, $0.95 cache-miss input, $4.00 output per 1M tokens, 262,144-token context.
- `kimi-k2.6`: $0.16 / $0.95 / $4.00 per 1M tokens and 262,144-token context.
- `moonshot-v1-128k`: 131,072 tokens, $2.00 input and $5.00 output per 1M tokens.
- `deepseek-v4-flash`: $0.14 cache-miss input and $0.28 output per 1M tokens, useful as a cost baseline.
What changed recently
- K2.7 Code and HighSpeed are the coding entries; Kimi Code is the workflow entry.
- Kimi API public pages show token billing, recharge, limits, and promotions rather than a fixed monthly API plan.
Explanation
Use K2.7 Code for coding and repo-agent work.
Use K2.6 for general multimodal and agent tasks.
Use K2.5 when Kimi cost matters most.
Use DeepSeek-V4-Flash as the API cost baseline in comparisons.
Kimi model and pricing quick reference
Read this table before quoting Kimi API cost.
| Model | Context | Input cache hit | Input cache miss | Output | Best fit |
|---|---|---|---|---|---|
| `kimi-k2.7-code` | 262,144 tokens | $0.19 | $0.95 | $4.00 | coding model |
| `kimi-k2.7-code-highspeed` | 262,144 tokens | $0.38 | $1.90 | $8.00 | high-speed coding |
| `kimi-k2.6` | 262,144 tokens | $0.16 | $0.95 | $4.00 | general multimodal / agent |
| `kimi-k2.5` | 262,144 tokens | $0.10 | $0.60 | $3.00 | lower-cost Kimi |
| `moonshot-v1-128k` | 131,072 tokens | n/a | $2.00 input | $5.00 output | V1 reference |
| `deepseek-v4-flash` | 1M | $0.0028 | $0.14 | $0.28 | API cost baseline |
How to verify the answer
Official pages used for model names, prices, context, and Kimi Code facts.
Tools / Examples
- 50k + 10k sample — `kimi-k2.7-code` is about $0.0875 cache miss; DeepSeek-V4-Flash is about $0.0098 cache miss.
- Kimi Code pilot — Use a clean branch, one README/test/script task, command output, touched files, and failure notes.
Evidence timeline
Sources
- Kimi model list
- Kimi K2.7 Code pricing
- Kimi K2.6 pricing
- Kimi K2.5 pricing
- Moonshot V1 pricing
- Kimi Code GitHub
- DeepSeek models pricing
FAQ
Does Kimi API have a monthly plan here?
The public API pages show token usage pricing, cache behavior, recharge, limits, and promotions, not a fixed monthly API plan.
Which Kimi model starts a coding workflow pilot?
Start with `kimi-k2.7-code`; use HighSpeed when latency is worth the higher unit price.
Related
- Chinese AI models list
- Chinese open-source AI models
- China AI updates
- China AI news sources in English
- Kimi vs DeepSeek for builders
- WHILE (topic)
- TIME (topic)
- Prompt injection and LLM security basics
- Prompting vs RAG vs fine-tuning (decision guide)
- Paradigm (topic)
- NVIDIA (topic)
- Meanwhile (topic)
- Kimi / Moonshot AI model and pricing quick reference
- Inference (topic)
- HAS (topic)
- Generation (topic)
- Development (topic)
Go deeper
- Chinese AI models list
- Chinese open-source AI models
- China AI updates
- China AI news sources in English
Last updated: 2026-07-22 · Policy: Editorial standards · Methodology