Kimi K2.5
Kimi K2.5Alternatives?
CheaperDeepSeek V4.1 Flash 1017maxOpenwins 53%-56%$0.30 / $1.20Pass back reasoning_content · Chat Completions+4
BetterGPT-5.5xhighwins 60%+838%$2.63Prompts with >272K input tokensOpenDeepSeek V4 Pro 0813maxwins 53%+65%$1.32 / $3.96Memory (4-bit): 616 GB → 990 GB+6
CheaperGemini 3 Flash Previewunstatedsame-6%$0.50 / $3
CheaperStep 3.7 FlashunstatedOpensame-64%$0.20 / $1.15DeepSeek V4.1 Flash 1017maxOpenwins 53%-56%$0.30 / $1.20Pass back reasoning_content · Chat Completions+4Gemini 3 Flash Previewunstatedsame-6%$0.50 / $3Step 3.7 FlashunstatedOpensame-64%$0.20 / $1.15
Qwen 3.5 35B A3BunstatedOpensame-43%$0.25 / $2Qwen 3.5 27BunstatedOpensame-31%$0.30 / $2.40GPT-5.5xhighwins 60%+838%$2.63Prompts with >272K input tokens
Claude Opus 4.8maxwins 60%+733%$4.08GPT-5.4 Proxhighwins 59%+5,525%$30 / $180No strict JSONKimi K3unstatedOpenwins 58%+400%$3 / $15Memory (4-bit): 616 GB → 1,668 GB+4Gemini 3.1 Pro Previewunstatedwins 57%+275%$2 / $12Cannot use both thinking controlsDeepSeek V4 Pro 0813maxOpenwins 53%+65%$1.32 / $3.96Memory (4-bit): 616 GB → 990 GB+6DeepSeek V4.1 Flash 1017maxOpenwins 53%-56%$0.30 / $1.20Pass back reasoning_content · Chat Completions+4Kimi K2.6unstatedOpenwins 52%+43%$0.95 / $4
GLM 5.2unstatedOpensame+79%$1.40 / $4.40No image inputStep 3.7 FlashunstatedOpensame-64%$0.20 / $1.15Kimi K3unstatedOpenwins 58%—$3 / $15Memory (4-bit): 616 GB → 1,668 GB+4Kimi K2.6unstatedOpenwins 52%—$0.95 / $4Kimi K2 ThinkingunstatedOpensame—$0.47 / $2No image inputKimi K3.1 Pron/a—$2 / $10Kimi K3.1 Pro Highspeedn/a—$4 / $20Kimi K2.7 CodethinkingOpenn/a—$0.54Instant mode is not supported · Self-hosted / vLLM / SGLangKimi K2 0905non-thinkingOpenn/a—$0.40 / $2No image inputKimi K2 0711non-thinkingOpenn/a—$0.50 / $2Context: 256K → 128K+2All Moonshot AI models →Run it yourself
estimateHistory
5 versionsBenchmarks
in Chat & KnowledgeFAQ
How much does Kimi K2.5 cost?›
$0.60 per 1M input tokens and $3 per 1M output tokens; cached input $0.10 per 1M; cheapest via DeepInfra at $0.45 / $2.25.
Is Kimi K2.5 free to use?›
There is no free API; the cheapest paid price is $0.45 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Kimi K2.5, and what is it best at?›
#76 of 334 in Chat & Knowledge, #56 of 97 in Coding, #96 of 175 in Agents & Tool Use, #51 of 195 in Writing & Language, #47 of 124 in Vision & Documents. Strongest at: Tool calling, Reasoning, Science, Long context, Instructions.
Is Kimi K2.5 better than DeepSeek V3.2 Speciale?›
On MathArena · Apex, Kimi K2.5 scores 8.9 vs 9.4 — about the same.
How can I use Kimi K2.5?›
Through the API of AWS Bedrock, Amazon Bedrock, AtlasCloud, Azure AI Foundry, DeepInfra and 3 more.
What is the context window of Kimi K2.5?›
256K tokens, with up to 262K output tokens.
When was Kimi K2.5 released? Is there a newer version?›
Released 2026-01-27. It has been superseded by Kimi K2.6.
Is Kimi K2.5 open source?›
Yes, the weights are open (modified-mit licence).
Can I run Kimi K2.5 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Kimi K2.5 read images?›
It accepts text and images as input.
What is Kimi K2.5?›
Kimi K2.5 is an open-weights language model from Moonshot AI, released Jan 2026, the successor to Kimi K2 0905.
What are the best alternatives to Kimi K2.5?›
DeepSeek V4.1 Flash 1017 (cheaper), GPT-5.5 (better), DeepSeek V4 Pro 0813 (open weights).