Kimi
Moonshot AI's Kimi family of large language models, including open-weight releases like Kimi K3.
HN commenters use Kimi K3 as a daily driver for coding, planning and reviews,12345 often comparing it to Opus-tier models,678 and value that it takes on security work that US models refuse.9101112 The most common complaint is high token usage and cost,13141516 and some say quality has declined since version 2.7.1718
Why people recommend it
Frontier-level quality · 3
Several commenters find it comparable to Opus or Claude and see little gap with frontier models.
“it is the first of the chinese models that actually feels comparable to Opus-tier models”
“Just tried K3 out for the first time today and it's a legitimate threat.”
Show 1 more
“I would generally consider the SWE capabilities of Kimi to be on par with Claude”
Coding daily driver · 3
Commenters have replaced Anthropic or other frontier models with it for everyday coding.
“Now it's my daily driver and I have zero complaints.”
“Except a month ago I started testing Kimi K3 and omp and I never went back.”
Show 1 more
“I have completed stopped using Anthropic since Kimi 3 released, and only ever K3 and Codex nowadays”
Fewer refusals · 3
It completes security and exploit work that GPT, Gemini and Opus refused.
“No rejections, they did find the same problems on my project that I used for testing.”
“I just switch the model to Kimi K3 to finish up making the PoC”
Show 1 more
“Switched to K3 and continued my work. And it worked really well for what I was doing.”
Planning and review · 3
Used for adversarial plan reviews, research, critique and finding bugs in old code.
“Combined with Kimi K3 for adversarial plan reviews - you get robust plans .”
“now Kimi K3 offers a great planning assistant for more advanced coding tasks”
Show 1 more
“GLM-5.3 for coding or Kimi K.3 for research and critique.”
Open weights · 3
Commenters value that it is an open model they can run through multiple providers.
“models like Kimi K3 are right there, open and free to use, as an alternative to the closed models”
“I have been using GLM and Kimi models the last few months mixed with the latest Anthropic models”
Show 1 more
“Check out Kimi, DeepSeek, and GLM series models in particular”
Writing style · 1
One commenter praises its writing style enough to use it to clean up verbose comments.
“Kimi K3 has been my go-to model for its writing style”
Caveats and dissent
Cost and tokens · 3
Some find it expensive per unit of work, and one hit a plan's 5-hour budget in minutes.
“Kimi is by far the most expensive in terms of tokens and $/work done.”
“Kimi K3 is smarter, though.”
Show 1 more
“Kimi didn't even finish the scan, ran out of the 5hr budget in less than 10 minutes. Unusable, unfortunately.”
Quality decline · 2
Former heavy users say it got worse starting with version 2.7.
“since 2.7 Kimi has gone way downhill”
Inconsistent output · 1
One commenter finds output quality varies more than with Anthropic or OpenAI models.
“they're incredibly inconsistent and the variance in output quality is much higher”
Slow responses · 2
It can be slow and take a long time to think, especially after the K3 launch.
“It's a good model otherwise.”
“I have been using Kimi for months.”
Data training · 1
One commenter warns that the default terms allow training on user data.
“Their EULA is “we use your data and train on it unless you negotiate something with us privately””
Usage caps · 1
The $19 plan's weekly cap was hit in a day and a half.
“it is the first of the chinese models that actually feels comparable to Opus-tier models”
Often compared with
2 moved from GLM to Kimi
4 moved from Kimi to GLM
2 moved from Kimi to DeepSeek
1 moved from Claude Opus to Kimi
6 moved from Claude to Kimi
2 moved from Claude Code to Kimi
1 moved from DeepSeek V4 Flash to Kimi
2 moved from Kimi to DeepSeek V4 Flash
Who recommends it in their own words
Threads where it came up
- Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA 7 mentions
- Kimi K2.7-Code: open-source coding model with better token efficiency 6 mentions
- Kimi K3.1 Pro and K3.1 Pro Highspeed 5 mentions
- Kimi K3 is now live 5 mentions
- GLM-5.3: Frontier Coding with Emergent Cyber Capabilities 4 mentions
- DeepSeek V4 Flash 0731 4 mentions
Also written as Kimi K3, Kimi K2.6, Kimi K2.5, Kimi K2. Data: kimi.json. Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 51
“while GLM, Kimi and others stay the same”
“Now it's my daily driver and I have zero complaints.”
“I used Kimi K3.1 Pro to figure it out”
“I am entirely satisfied when using my Opencode Go plan with open models like Deepseek Flash/Pro, Kimi and MiMo.”
“Kimi is by far the most expensive in terms of tokens and $/work done.”
“Found it functional but boring. Preferred Qwen.”
“glm and kimi k.3, expensive, slow, ultimately not that much better than deepseek flash for this task.”
“now this is just a better model with the same price”
“Still burning through the 5-hour rate limit way too quickly, though. Output quality seems solid enough”
“Kimi K3 has been my go-to model for its writing style”
“Kimi didn't even finish the scan, ran out of the 5hr budget in less than 10 minutes. Unusable, unfortunately.”
“models like Kimi K3 are right there, open and free to use, as an alternative to the closed models”
“compile the prompt for the agentic coding tool using Kimi K3”
“1B on the $40 plan where I used more Kimi k3”
“Kimi K2.5, 2.7 and Deepseek v4 were among my favorites”
“Kimi K2.5, 2.7 and Deepseek v4 were among my favorites”
“GLM-5.3 for coding or Kimi K.3 for research and critique.”
“I use MiMov2.6Pro, DeepSeekv4.1Flash, GLM5.3, Hy4, Qwen3.8 and KimiK3 at home.”
“qwen3.8, kimi3, kimi2.7, GLM-5.3 are all good families I use in my coding team”
“qwen3.8, kimi3, kimi2.7, GLM-5.3 are all good families I use in my coding team”
“since 2.7 Kimi has gone way downhill”
“this was with Kimi which is one of the open models”
“I would generally consider the SWE capabilities of Kimi to be on par with Claude”
“The Chinese models are simply _excellent_”
“Kimi K2.5 before that”
“Kimi and GLM feel extremely coding oriented. I use them for code reviews basically.”
“Kimmi with whisper.”
“For a long time Kimi K3 was the best model for this work”
“or kimi 3 on CPU”
“offering Kimi k3 performance for like half the price”
“vs Kimi 2.7 that I've been using in the cloud version”
“Allow me to recommend switching to Kimi K3 / GLM 5.3 / DeepSeek.”
“GLM-5.3, GLM-5.3-Flash and Kimi K3 are also fine, but slower, more expensive”
“even if I managed to switch to Kimi K3”
“I just switch the model to Kimi K3 to finish up making the PoC”
“I use something top-tier like Kimi K3”
“Combined with Kimi K3 for adversarial plan reviews - you get robust plans .”
“Kimi 2.7 paired with Gemini 3.5/4.6/3.7 flash has been my implementer”
“I use Kimi K3 and GLM 5.3 to find old bugs and improve old code”
“I find it hard to distinguish between the outputs of GPT/Claude/Kimi/GLM recently”
“Except a month ago I started testing Kimi K3 and omp and I never went back.”
“Fwiw I love K3 and use it as a daily driver.”
“So, I switched to Kimi K3 and DeepSeek for API testing just because it's so much cheaper.”
“Kimi K3 is smarter, though.”
“Kimi 3 was super expensive as spec model”
“I was using Kimi a lot but it just nose dived.”
“Just tried K3 out for the first time today and it's a legitimate threat.”
“it is the first of the chinese models that actually feels comparable to Opus-tier models”
“No rejections, they did find the same problems on my project that I used for testing.”
“now Kimi K3 offers a great planning assistant for more advanced coding tasks”
“Switched to K3 and continued my work. And it worked really well for what I was doing.”