How much does Kimi K2.5 cost?

Kimi K2.5 costs $0.49/M input tokens and $2.50/M output tokens, with separate rates for Cache Read at $0.20/M tokens.

What is the context length of Kimi K2.5?

Kimi K2.5 has a 262,144 token context window. It supports up to 262,144 completion tokens.

Does Kimi K2.5 support tool calling and structured outputs?

Yes. Kimi K2.5 accepts tools and tool_choice for function calling. It also supports structured outputs via a JSON schema in response_format.

What inputs and outputs does Kimi K2.5 support?

Kimi K2.5 accepts text and images as input and returns text.

Which providers serve Kimi K2.5?

Kimi K2.5 is served by 5 providers on OpenRouter: AtlasCloud, SiliconFlow, Venice, NovitaAI, and Amazon Bedrock. Requests are routed to the best available provider, with automatic failover to the others, and you can pin or exclude providers with provider routing.

When was Kimi K2.5 released?

Kimi K2.5 was released on January 27, 2026.

MoonshotAI: Kimi K2.5

moonshotai/kimi-k2.5

Model weights

Compare

Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm paradigm. Built on Kimi K2 with continued pretraining over approximately 15T mixed visual and text tokens, it delivers strong performance in general reasoning, visual coding, and agentic tool-calling.

Modalities

In / out price

$0.49 / $2.50per 1M

Context

262K

Released

Jan 27, 2026

Providers

Different companies host the same model. OpenRouter routes your request to one of them based on the routing mode you pick: Balanced (price + speed), Nitro (fastest), or Floor (cheapest).

Latency / throughput


	$0.50	$2.50	$0.15	1.70s	26 tok/s	100.00%
5% off	$0.56$0.532	$3.50$3.325	$0.22$0.209	1.16s	72 tok/s	100.00%
5% off	$0.60$0.57	$3.00$2.85	$0.10$0.095	1.22s	32 tok/s	98.85%
	$0.60	$3.00	--	2.11s	35 tok/s	99.89%
	$0.49	$2.50	$0.20	0.80s	49 tok/s	98.44%

Input $0.50Output $2.50
26 tok/s100.00% uptime
Input $0.532Output $3.325
5% off
72 tok/s100.00% uptime
Input $0.57Output $2.85
5% off
32 tok/s98.85% uptime
Input $0.60Output $3.00
35 tok/s99.89% uptime
Input $0.49Output $2.50
49 tok/s98.44% uptime

Pricing

The average price customers actually pay for this model, next to the prices providers post. Caching and discounts mean the price actually paid is often well below the listed one.

Weighted average

Input

$0.2989/M tokens

Output

$2.95/M tokens

Kimi K2.5 — Price history

Price history

No effective price data for this selection

$0.246	$2.50	72.7%	31.0%
$0.255	$2.50	81.0%	7.2%
$0.268	$2.85	63.6%	19.5%
$0.336	$3.32	60.6%	38.4%
$0.600	$3.00	0.0%	3.9%

Effective in $0.246Effective out $2.50
Token share
31.0%
Effective in $0.255Effective out $2.50
Token share
7.2%
Effective in $0.268Effective out $2.85
Token share
19.5%
Effective in $0.336Effective out $3.32
Token share
38.4%
Effective in $0.600Effective out $3.00
Token share
3.9%

Performance

Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).

Throughput

72tok/s

P50, best across providers

Latency

0.80s

P50, best provider

Latency / throughput

AutoExacto benchmarks

GPQA DiamondTAU-BenchNovitaAI87.6%--SiliconFlow90.4%72.6%Venice89.6%68.9%auto-routing84.3%67.2%Amazon Bedrock80.0%57.1%

Uptime

Uptime is the percentage of the past 3 days that at least one provider was responding to requests. Availability is the percentage of time that inference was successfully served. OpenRouter continuously monitors and uses the next-best provider when one returns an error.

Uptime (3d)

100.00%

Availability (3d)

99.65%

Availability over the last 3 days

Last 72 hours

Availability 99.65%

3 Days Ago2 Days AgoYesterdayNow

Availability over the last 24 hours

OpenRouter Availability

99.63%

Without Routing

96.28%

When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it. You can access per-provider uptime data programmatically through the Endpoints API. Learn more about our load balancing and provider routing options.

Benchmarks

Scores on standardized evaluations and this model’s own outputs on shared benchmark prompts. Higher percentages are better — and each rank shows where this model lands among the models evaluated on that source.

Benchmark score summary for MoonshotAI: Kimi K2.5 (Artificial Analysis, Design Arena and Vals AI)
Source	Benchmark	Score
Artificial Analysis	Kimi K2.5 (Non-reasoning) GPQA Diamond	78.9%
Artificial Analysis	Kimi K2.5 (Non-reasoning) HLE	13.2%
Artificial Analysis	Kimi K2.5 (Non-reasoning) IFBench	43.7%
Artificial Analysis	Kimi K2.5 (Non-reasoning) τ²-Bench Telecom	81.3%
Artificial Analysis	Kimi K2.5 (Non-reasoning) AA-LCR	67.3%
Artificial Analysis	Kimi K2.5 (Non-reasoning) CritPt	0.6%
Artificial Analysis	Kimi K2.5 (Non-reasoning) Terminal-Bench Hard	18.9%
Artificial Analysis	Kimi K2.5 (Reasoning) Coding Index	46.8
Artificial Analysis	Kimi K2.5 (Reasoning) GPQA Diamond	87.9%
Artificial Analysis	Kimi K2.5 (Reasoning) HLE	30.7%
Artificial Analysis	Kimi K2.5 (Reasoning) IFBench	70.2%
Artificial Analysis	Kimi K2.5 (Reasoning) τ²-Bench Telecom	95.9%
Artificial Analysis	Kimi K2.5 (Reasoning) AA-LCR	78.0%
Artificial Analysis	Kimi K2.5 (Reasoning) τ-Bench Banking	14.2%
Artificial Analysis	Kimi K2.5 (Reasoning) GDPval-AA	17.3%
Artificial Analysis	Kimi K2.5 (Reasoning) CritPt	3.1%
Artificial Analysis	Kimi K2.5 (Reasoning) Terminal-Bench Hard	34.8%
Artificial Analysis	Kimi K2.5 (Reasoning) Terminal-Bench 2.1	45.7%
Design Arena	Kimi K2.5 (Thinking) Agents Arena Androidnative Elo	1105
Design Arena	Kimi K2.5 (Thinking) Agents Arena Full Stack Elo	1088
Design Arena	Kimi K2.5 (Thinking) Agents Arena Godotgamedev Elo	1219
Design Arena	Kimi K2.5 (Thinking) Agents Arena Mobile Apps Elo	1139
Design Arena	Kimi K2.5 (Thinking) Agents Arena Webapps Elo	1114
Design Arena	Kimi K2.5 (Thinking) Models Arena 3D Elo	1184
Design Arena	Kimi K2.5 (Thinking) Models Arena Asciiart Elo	1160
Design Arena	Kimi K2.5 (Thinking) Models Arena Code Categories Elo	1226
Design Arena	Kimi K2.5 (Thinking) Models Arena Data Visualization Elo	1216
Design Arena	Kimi K2.5 (Thinking) Models Arena Game Development Elo	1196
Design Arena	Kimi K2.5 (Thinking) Models Arena SVG Elo	1139
Design Arena	Kimi K2.5 (Thinking) Models Arena UI Component Elo	1227
Design Arena	Kimi K2.5 (Thinking) Models Arena Website Elo	1240
Vals AI	Kimi K2.5 AIME	95.6%
Vals AI	Kimi K2.5 CaseLaw v2	58.7%
Vals AI	Kimi K2.5 Code Migration	7.0%
Vals AI	Kimi K2.5 CorpFin v2	68.3%
Vals AI	Kimi K2.5 EMB	28.5%
Vals AI	Kimi K2.5 Finance Agent (v2)	35.8%
Vals AI	Kimi K2.5 GPQA Diamond	84.1%
Vals AI	Kimi K2.5 Harvey's Legal Agent Benchmark v1	0.0%
Vals AI	Kimi K2.5 IOI v1	17.7%
Vals AI	Kimi K2.5 LiveCodeBench	83.9%
Vals AI	Kimi K2.5 Legal Research Bench	15.9%
Vals AI	Kimi K2.5 MedCode	39.3%
Vals AI	Kimi K2.5 MedQA	94.4%
Vals AI	Kimi K2.5 MedScribe	76.4%
Vals AI	Kimi K2.5 MMLU Pro	85.9%
Vals AI	Kimi K2.5 MMMU Pro	84.3%
Vals AI	Kimi K2.5 MortgageTax	66.5%
Vals AI	Kimi K2.5 Vals RSI Index	6.8%
Vals AI	Kimi K2.5 SAGE v1	49.9%
Vals AI	Kimi K2.5 SWE-bench	70.0%
Vals AI	Kimi K2.5 TaxEval v2	74.2%
Vals AI	Kimi K2.5 Terminal-Bench 2.1	41.9%
Vals AI	Kimi K2.5 Vals Index	26.3%
Vals AI	Kimi K2.5 Vals Multimodal Index	52.0%
Vals AI	Kimi K2.5 Vibe Code Bench v1.1	17.5%

Apps

Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.

OpenClaw

OpenClaw is an open-source AI agent that connects to your messaging apps and takes real actions on your behalf, from running commands and browsing the web to managing files and sending emails.

14.1Btokens

Janitor AI

Janitor AI is a chatbot platform where users create and chat with custom AI characters for interactive roleplay, storytelling, and immersive fiction.

10.9Btokens

Hermes Agent

Hermes Agent is an open-source, self-improving AI agent by Nous Research that runs persistently with memory across sessions, and builds reusable skills from experience. It comes with 40+ built-in tools, including web search, browser automation, and vision, plus scheduled automations and subagents.

6.82Btokens

SillyTavern

SillyTavern is the LLM frontend for power users, a chat interface that connects to any model and gives deep control through character creation, roleplay, and prompt customization.

1.72Btokens

predictionTrading

new

1.52Btokens

Rest of OpenRouter

534M

Top 5 Apps

308M

Not currently available through EU in-region routing. In-region routing docs

More models from moonshotai

Kimi K3.1 Pro

Kimi K3.1 Pro is Moonshot AI's flagship open-weight multimodal reasoning model and a post-trained successor to Kimi K3, with a 1M-token context window. It is built for long-horizon coding, 3D creation, financial research, and end-to-end agentic knowledge work, with native image and video understanding.

Text1.0M context$0.2375 / $8.411

Kimi K3.1 Pro (batch)

Text1.0M context$0.70 / $6

Kimi K3

Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at navigating large repositories, using tools, debugging, and iterating against images, logs, tests, and runtime feedback. Its architecture uses KDA and Attention Residuals for computational efficiency.

Text1.0M context$0.1824 / $11.01

Kimi K3 (batch)

Text1.0M context$2.28 / $11.40

Kimi K2.7 Code

MoonshotAI: Kimi K2.7 Code is a coding-focused model in Moonshot AI's Kimi K2 family, built to complete end-to-end programming tasks reliably over long contexts. It uses a native multimodal mixture-of-experts architecture that accepts text and image input, and it always operates in a thinking mode, preserving full reasoning content across multi-turn conversations. With a 256K-token context window, it targets long-horizon coding, agentic task decomposition, and multi-turn dialogue. The model activates 32B parameters out of roughly 1T total.

Text262K context$0.60 / $3

Kimi Latest

This model always redirects to the latest model in the Kimi family.

Text262K context

Kimi K2.6

Kimi K2.6 is Moonshot AI's next-generation multimodal model, designed for long-horizon coding, coding-driven UI/UX generation, and multi-agent orchestration. It handles complex end-to-end coding tasks across Python, Rust, and Go, and can convert prompts and visual inputs into production-ready interfaces. Its agent swarm architecture scales to hundreds of parallel sub-agents for autonomous task decomposition - delivering documents, websites, and spreadsheets in a single run without human oversight.

Text262K context$0.3895 / $1.64

Kimi K2.5

Text262K context$0.49 / $2.50

Kimi K2 Thinking

Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter Mixture-of-Experts (MoE) architecture introduced in Kimi K2, it activates 32 billion parameters per forward pass and supports 256 k-token context windows. The model is optimized for persistent step-by-step thought, dynamic tool invocation, and complex reasoning workflows that span hundreds of turns. It interleaves step-by-step reasoning with tool use, enabling autonomous research, coding, and writing that can persist for hundreds of sequential actions without drift.

It sets new open-source benchmarks on HLE, BrowseComp, SWE-Multilingual, and LiveCodeBench, while maintaining stable multi-agent behavior through 200–300 tool calls. Built on a large-scale MoE architecture with MuonClip optimization, it combines strong reasoning depth with high inference efficiency for demanding agentic and analytical tasks.

Text262K context$0.60 / $2.50

Kimi K2 0905

Kimi K2 0905 is the September update of Kimi K2 0711. It is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass. It supports long-context inference up to 256k tokens, extended from the previous 128k.

This update improves agentic coding with higher accuracy and better generalization across scaffolds, and enhances frontend coding with more aesthetic and functional outputs for web, 3D, and related tasks. Kimi K2 is optimized for agentic capabilities, including advanced tool use, reasoning, and code synthesis. It excels across coding (LiveCodeBench, SWE-bench), reasoning (ZebraLogic, GPQA), and tool-use (Tau2, AceBench) benchmarks. The model is trained with a novel stack incorporating the MuonClip optimizer for stable large-scale MoE training.

Text262K context$0.60 / $2.50

Kimi K2 0711

Kimi K2 Instruct is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass. It is optimized for agentic capabilities, including advanced tool use, reasoning, and code synthesis. Kimi K2 excels across a broad range of benchmarks, particularly in coding (LiveCodeBench, SWE-bench), reasoning (ZebraLogic, GPQA), and tool-use (Tau2, AceBench) tasks. It supports long-context inference up to 128K tokens and is designed with a novel training stack that includes the MuonClip optimizer for stable large-scale MoE training.

Text131K context$0.57 / $2.30

Kimi Dev 72B

Kimi-Dev-72B is an open-source large language model fine-tuned for software engineering and issue resolution tasks. Based on Qwen2.5-72B, it is optimized using large-scale reinforcement learning that applies code patches in real repositories and validates them via full test suite execution—rewarding only correct, robust completions. The model achieves 60.4% on SWE-bench Verified, setting a new benchmark among open-source models for software bug fixing and code reasoning.

Text131K context

Kimi VL A3B Thinking

Kimi-VL is a lightweight Mixture-of-Experts vision-language model that activates only 2.8B parameters per step while delivering strong performance on multimodal reasoning and long-context tasks. The Kimi-VL-A3B-Thinking variant, fine-tuned with chain-of-thought and reinforcement learning, excels in math and visual reasoning benchmarks like MathVision, MMMU, and MathVista, rivaling much larger models such as Qwen2.5-VL-7B and Gemma-3-12B. It supports 128K context and high-resolution input via its MoonViT encoder.

Text131K context

Moonlight 16B A3B Instruct

Moonlight-16B-A3B-Instruct is a 16B-parameter Mixture-of-Experts (MoE) language model developed by Moonshot AI. It is optimized for instruction-following tasks with 3B activated parameters per inference. The model advances the Pareto frontier in performance per FLOP across English, coding, math, and Chinese benchmarks. It outperforms comparable models like Llama3-3B and Deepseek-v2-Lite while maintaining efficient deployment capabilities through Hugging Face integration and compatibility with popular inference engines like vLLM12.

Text8K context

MoonshotAI: Kimi K2.5

Providers