Kimi K3
Kimi K3Alternatives?
CheaperGrok 4.6high+10%-26%$1.48Context: 1M → 500K
BetterQwen3.8 27BxhighOpen+4%-50%$1.01Context: 1M → 262K+5
OpenGLM 5.3max+9%same$2.01No image input+4Lower effortSame modellowOpen-9%-43%$1.15
CheaperMuse Spark 1.3max+10%-20%$1.60Only auto tool_choice is supported+5Grok 4.6high+10%-26%$1.48Context: 1M → 500KMuse Spark 1.3max+10%-20%$1.60Only auto tool_choice is supported+5Qwen3.8 27BxhighOpen+4%-50%$1.01Context: 1M → 262K+5GLM 5.3 FlashmaxOpen+3%-86%$0.28Defaults to max · Self-hosted+4
Gemini 3.8 Flashmediumsame-53%*$0.93Sampling parameters are ignored+1Qwen3.8 27BxhighOpen+4%-50%$1.01Context: 1M → 262K+5Grok 4.6high+10%-26%$1.48Context: 1M → 500KMuse Spark 1.3max+10%-20%$1.60Only auto tool_choice is supported+5GLM 5.3maxOpen+9%same$2.01No image input+4Qwen3.8 2.4T A95BthinkingOpen+7%+8%$2.16Context: 1M → 262K+6GLM 5.3maxOpen+9%same$2.01No image input+4Qwen3.8 2.4T A95BthinkingOpen+7%+8%$2.16Context: 1M → 262K+6Qwen3.8 27BxhighOpen+4%-50%$1.01Context: 1M → 262K+5GLM 5.3 FlashmaxOpen+3%-86%$0.28Defaults to max · Self-hosted+4Qwen3.8 FlashthinkingOpensame-81%$0.37Context: 1M → 262K+6Kimi K2.6thinkingOpen-49%-60%$0.80Context: 1M → 256KKimi K2.7 CodethinkingOpen-56%-73%$0.54Context: 1M → 256K+1Kimi K2.5thinkingOpen-69%—$0.60 / $3Context: 1M → 256KKimi K3.1 Pron/a—$2 / $10Kimi K3.1 Pro Highspeedn/a—$4 / $20Kimi K2 ThinkingthinkingOpenn/a—$0.47 / $2Context: 1M → 256K+1Kimi K2 0905non-thinkingOpenn/a—$0.40 / $2Context: 1M → 256K+1Kimi K2 0711non-thinkingOpenn/a—$0.50 / $2Context: 1M → 128K+2All Moonshot AI models →Run it yourself
estimateHistory
5 versionsBenchmarks
in Agents & Tool UseWatch out 4
Sampling parameters are fixed›
temperature, top_p, n and penalty values are fixed; omit them from requests. Source ↗
No public image URLs›
Vision input rejects public image URLs; use base64 or ms:// file references in a content array. Source ↗
Web search is being updated›
Web search is not recommended for near-term production workflows; avoid depending on it in production. Source ↗
Thinking mode is always enabled›
K3 always has thinking enabled; control effort with the top-level reasoning_effort field using low, high or max. Source ↗
FAQ
How much does Kimi K3 cost?›
$3 per 1M input tokens and $15 per 1M output tokens; cached input $0.30 per 1M; cheapest via Relace at $0.80 / $3. A typical benchmark task costs about $2.00 at max effort.
Is Kimi K3 free to use?›
There is no free API; the cheapest paid price is $0.80 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Kimi K3, and what is it best at?›
#19 of 334 in Chat & Knowledge, #34 of 132 in Coding, #23 of 133 in Agents & Tool Use, #17 of 195 in Writing & Language, #16 of 124 in Vision & Documents. Strongest at: Agents & work tasks, Reasoning, Science, Long context, Knowledge, Vision.
Is Kimi K3 better than Kimi K2.6?›
On τ³-Banking · AA, Kimi K3 scores 46 vs 23.3 — higher. It is 150% more expensive per task.
How can I use Kimi K3?›
Through the API of AkashML, Alibaba, BaseTen, Chutes, DeepInfra and 13 more.
What is the context window of Kimi K3?›
1M tokens, with up to 944K output tokens.
When was Kimi K3 released? Is there a newer version?›
Released 2026-07-16. It is the current version in its line.
Is Kimi K3 open source?›
Yes, the weights are open (kimi-k3 licence).
Can I run Kimi K3 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Kimi K3 read images?›
It accepts text and images, video as input.
What is Kimi K3?›
Kimi K3 is an open-weights language model from Moonshot AI, released Jul 2026, the successor to Kimi K2.6.
What are the best alternatives to Kimi K3?›
Grok 4.6 (cheaper), Qwen3.8 27B (better), GLM 5.3 (open weights).