AI chatbots and agents , by Moonshot AI , since 2023
Kimi
Moonshot AI's chatbot built on the open-weight Kimi K2 models, strong for long-context coding
- WEB
- API
- MOBILE
What Kimi does
Kimi is a Chinese AI chatbot from Moonshot AI built around the open-weight Kimi K2 model family, positioned for agentic coding and knowledge work with very large context windows. It offers a free consumer tier plus paid membership tiers for heavier agent/coding usage, and a separately-billed API popular for its low per-token cost relative to Western models.
What you get: Kimi K2 open-weight model, Very large context window, Kimi Code agentic coding, Low-cost API with cache discounts, Free unlimited-ish basic chat, Plugins and scheduled tasks.
Typical jobs: free long context ai chatbot; cheap ai coding agent alternative; chinese ai chatbot for developers; kimi vs deepseek comparison; ai chatbot for large document analysis.
Worth knowing: Official pricing page blocked automated fetching during verification; tier names/prices reconstructed from multiple independent 2026 pricing trackers that agree closely with each other.
What users complain about: Privacy concerns about Chinese data handling appear in nearly every Kimi discussion thread, especially for hosted API use. (Reddit threads on Kimi (via aitooldiscovery roundup))
Pricing and plan limits
Verified August 2026| Plan | Monthly | Annual (per mo) | Limits |
|---|---|---|---|
| Free (Adagio) | Free | Free | Unlimited basic chat, file uploads and web search at standard rate limits |
| Allegro | $19 | $19 | Higher usage caps for agent and coding tasks |
| Top tier | $199 | $159 | Highest usage tier for heavy agent/coding workloads |
Where it wins
- Free tier is genuinely usable for everyday chat, not a crippled trial
- Very large context window handles long documents/codebases better than most competitors at this price
- API pricing (roughly $0.15-0.60 per million input tokens) is far below GPT/Claude equivalents
- Kimi Code is a credible low-cost alternative to Claude Code/Copilot for agentic coding
Where it falls short
- Data is processed on Chinese company infrastructure, a recurring privacy concern in Reddit threads
- Loses coherence on very large context ('lost in the middle') when a huge repo is dumped in one go
- Tends to over-explain even simple answers, adding friction for users who want terse output
- Official pricing page is hard to access programmatically and changes tier names/prices frequently
How we scored it
Full methodology| Criterion | Weight | Score | Contribution |
|---|---|---|---|
| Capability | 30% | 7.9 | 2.37 |
| Value for money | 25% | 8.6 | 2.15 |
| Ease of use | 15% | 7.0 | 1.05 |
| Maturity | 15% | 6.0 | 0.90 |
| Support & docs | 10% | 5.0 | 0.50 |
| Momentum | 5% | 8.0 | 0.40 |
| Weighted total | 100% | 7.4 | 7.40 |
Alternatives to Kimi
Chosen by editors, not sponsors
Similar tools in AI chatbots and agents
OpenAI's general-purpose AI chatbot with voice, image, agent and coding modes
- WEB
- API
- MOBILE
- DESKTOP
- EXTENSION
Anthropic's conversational AI assistant known for long context, coding and careful reasoning
- WEB
- API
- MOBILE
- DESKTOP
- EXTENSION
Answer-engine chatbot that cites live web sources instead of just generating text
- WEB
- API
- MOBILE
- DESKTOP
- EXTENSION