Research-heavy workflows, long document analysis, users who need 128K+ context windows.
Simple chat use cases; less brand recognition than ChatGPT or Claude in Western markets.
Reviews from businesses like yours — matched by size and type. Available once you create a profile.
Complete profileRelated Tools
Community Reviews
Reviews from Reddit, Product Hunt, Hacker News & AlignAI members
It's a good opportunity for people to try kimi and others.
Hacker News • HackerNews • Use case: AI coding assistant (alternative to Claude Code)
I look at the output of Kimi and the costs of running inference on it that i can replicate, and it isn t that bad.
Hacker News • HackerNews • Use case: Running inference and evaluating output
kimi was in most of the cases better and my replacement for claude when i run out of tokens. Now im finding myself using glm more then kimi.
Hacker News • HackerNews • Use case: General AI assistant, replacement for Claude
kimi was best open source model for me. But it solved something that kimi k2.5 could not.
Hacker News • HackerNews • Use case: General AI tasks / problem-solving
I've been messing with using... Kimi... for reverse engineering... it's a ton of fun. Great because matching bytes is a scoring function that's easy for the models to understand and make progress on.
Hacker News • HackerNews • Use case: reverse engineering, matching bytes as a scoring function
We are currently evaling alternatives... Kimi, Chatgpt, Qwen are so far the best candidates.
Hacker News • HackerNews • Use case: evaluating as an alternative for coding tasks
it was not very good at the time.
Hacker News • HackerNews • Use case: Evaluating the model out of curiosity
I've been on Kimi now for 3 months. I rarely used Google in that time.
Hacker News • HackerNews • Use case: search engine
The key insight is that cheap models in consensus are more reliable than a single expensive model.
Hacker News • HackerNews • Use case: Part of a quorum/voting system for LLM responses and tool use
I tried this today. It s good - but it was significantly less focused and reliable than Opus 4.5... I found the model had to double-back to fix type issues, broken tests, etc, far more than Opus 4.5.
Hacker News • HackerNews • Use case: Implementing specifications for modifications to a Node.js/Express service
The other frontier models, and even open source Chinese ones like Kimi... understand perfectly fine what we mean by it.
Hacker News • HackerNews • Use case: Understanding nuanced instructions (like 'don't use academic language')
even if you use SOTA open source models like Kimi K2.5... they simply aren't as capable as the SOTA models from the frontier labs.
Hacker News • HackerNews • Use case: Considering as an open-source model for self-hosting an internal agent.
All of them but DS got it right. Implies Kimi K2.5 answered the problem correctly.
Hacker News • HackerNews • Use case: Testing its reasoning on a practical problem (whether to walk or drive a very short distance).
for taste one can also use Kimi K2.5.
Hacker News • HackerNews • Use case: Tasks where 'taste' is a priority
open source models (Kimi, Deepseek, Qwen) are getting better and better, and apple makes excellent hardware for local LLMs.
Hacker News • HackerNews • Use case: Referenced as an example of a high-quality open-source model suitable for local LLMs.
The Kimi K2 model from assistant are both excellent in helping find what I actually want to see.
Hacker News • HackerNews • Use case: Used via assistant to help find information
I reach for MiniMax first in my AI workflows, GLM for code tasks and Kimi K2.5 when deep English analysis is needed.
Hacker News • HackerNews • Use case: used when deep English analysis is needed
I have tried the same approach with Kimi K2.5... but I keep going back to Qwen3.
Hacker News • HackerNews • Use case: trying the same approach as with Qwen3
In the Kimi vendor verifier, Together has one of the highest tool call failure rates (compared to 0-2 for the official API...).
Hacker News • HackerNews • Use case: Benchmarking vendor performance via K2-Vendor-Verifier
Today we could run this beast of 1 trillion tokens, Kimi 2.5... It runs ~1 token/s with a 256gb ram and some paging memory. Currently under test, we are very excited to see the results.
Hacker News • HackerNews • Use case: Running a 1 trillion token model
Have You Used This Tool?
Sign in to share your experience with this tool.
Sign in