impulse

Kimi K2.5

Open weights

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.

API id kimi-k2.5Released 27 Jan 2026 (Jan 2026)
A newer model in this line is out: Kimi K3.1.
Azure AI Foundry retires it on 26 Jan 2027 (in 68 days). 9 hosts keep serving it.
Input / 1M
$0.60
at Moonshot AI
Output / 1M
$3.00
at Moonshot AI
Context window
262K
tokens
Max output
262K
tokens per reply
Hosts
10
33 listings
Released
Jan 2026
27 Jan 2026

Pricing

List prices at Moonshot AI, in US dollars per million tokens.

TierInput / 1MOutput / 1M
Standard$0.60$3.00
Cached input
83% off input
$0.10

What it costs a month

At Moonshot AI list prices, 30 days.

  • Chat assistant5,000 requests a day, 2K in / 0.5K out, cached$360
  • RAG over documents2,000 requests a day, 8K in / 0.6K out, cached$276
  • Coding agent1,000 requests a day, 40K in / 2K out, cached$480
  • Overnight batch20,000 requests a day, 4K in / 0.4K out$2,160
Use your own numbers

Where to run it

10 hosts serve Kimi K2.5. The comparison is on blended price, 3 parts input to 1 part output.

HostIn / out per 1M
kimi-k2.5
$0.60 / $3.00
DeepInfraInference
Kimi-K2.5
$0.45 / $2.25
−25% vs headline
kimi-k2.5
$0.60 / $3.00
Retires 26 Jan 2027
moonshotai.kimi-k2.5
$0.60 / $3.00
BasetenInference
Kimi-K2.5
$0.60 / $3.00
GMI CloudInference
Kimi-K2.5
$0.60 / $3.00
W&B InferenceInference
Kimi-K2.5
$0.60 / $3.00
Novita AIInference
kimi-k2.5
$0.60 / $3.00
Fireworks AIInference
kimi-k2p5
$0.60 / $3.00
ReplicateInference
kimi-k2.5
$0.72 / $3.60
+20% vs headline
Show 23 more listings: regions, snapshots and variants
Amazon Bedrock
Alias
moonshotai.kimi-k2.5
$0.60 / $3.00
us-east-1
moonshotai.kimi-k2.5
$0.60 / $3.00
us-east-2
moonshotai.kimi-k2.5
$0.60 / $3.00
us-west-2
moonshotai.kimi-k2.5
$0.60 / $3.00
Mantle
moonshotai.kimi-k2.5
$0.60 / $3.00
ap-southeast-2
moonshotai.kimi-k2.5
$0.62 / $3.09
+3% vs headline
ap-southeast-2 · Mantle
moonshotai.kimi-k2.5
$0.62 / $3.09
+3% vs headline
ap-northeast-1
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
ap-south-1
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
ap-southeast-3
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
eu-north-1
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
sa-east-1
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
eu-central-1
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
eu-south-1
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
eu-west-1
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
ap-northeast-1 · Mantle
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
ap-south-1 · Mantle
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
ap-southeast-3 · Mantle
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
eu-north-1 · Mantle
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
sa-east-1 · Mantle
moonshotai.kimi-k2.5
$0.72 / $3.60
+20% vs headline
eu-west-2
moonshotai.kimi-k2.5
$0.93 / $4.65
+55% vs headline
eu-west-2 · Mantle
moonshotai.kimi-k2.5
$0.93 / $4.65
+55% vs headline
Fireworks AI
Alias
kimi-k2p5
$0.60 / $3.00

Capabilities

Core

  • Streaming
  • Function calling
  • JSON mode
  • Structured outputs

Multimodal

  • Vision
  • Image input
  • Audio
  • Image generation
  • Video

Advanced

  • Reasoning
  • Tool calling
  • MCP
  • Prompt caching
  • Batch API
  • Embeddings
Takes in
TextImagesVideo
Produces
Text
Features
Video input
Open weights
License: Modified MIT License
Runs in about 2464.5 GB of GPU memory
Runtimes: SGLang, vLLM

Cheaper alternatives

Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.

Same model, cheaper host

  • DeepInfra
    Kimi-K2.5
    $0.45 / $2.25
    25% less

Smaller Moonshot AI models

No cheaper Moonshot AI model keeps its key capabilities.

Other frontier labs

  • Gemini 3.5 Flash Lite
    Google · on Google AI Studio
    $0.30 / $2.50
    Has every key capability
    29% less
  • GPT-5.4 Nano
    OpenAI · on OpenAI
    $0.20 / $1.25
    Has every key capability
    61% less
  • Claude Haiku 5.5
    Anthropic · on Anthropic
    $0.10 / $0.50
    Has every key capability
    83% less
  • GPT-6 Luna
    OpenAI · on OpenAI
    $0.10 / $0.50
    Has every key capability
    83% less

Open-weight picks

  • Kimi K2.7 Code
    Moonshot AI · on CLF AI Gateway
    $0.38 / $1.60
    Has every key capability
    43% less
  • Inkling Small
    Independent · on DeepInfra
    $0.45 / $1.20
    Has every key capability
    47% less
  • DeepSeek Flash
    DeepSeek · on DeepSeek
    $0.30 / $1.20
    Has every key capability
    56% less
  • Qwen3.7 Plus
    Alibaba Qwen · on Replicate
    $0.276 / $1.10
    Has every key capability
    60% less

Model line

Newer and older versions, at each lab's own price.

  1. UndatedKimi K3.1Latest$4.40 / $22.00
  2. Jul 2026Kimi K3$3.00 / $15.00
  3. UndatedKimi K2.7$0.95 / $4.00
  4. Apr 2026Kimi K2.6$0.95 / $4.00
  5. Jan 2026Kimi K2.5This page$0.60 / $3.00
  6. Jul 2025Kimi K2$0.55 / $2.20

Call it

Using Moonshot AI's API.

Python
from litellm import completion

response = completion(
    model="moonshot/kimi-k2.5",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Quick answers

How much does Kimi K2.5 cost?

On Moonshot AI, Kimi K2.5 costs $0.60 per million input tokens and $3.00 per million output tokens. Cached input costs $0.10 per million tokens. The cheapest host is DeepInfra, at $0.45 per million input tokens and $2.25 per million output tokens.

Which APIs offer Kimi K2.5?

Kimi K2.5 runs on 10 APIs: Moonshot AI, DeepInfra, Azure AI Foundry, Amazon Bedrock, Baseten, GMI Cloud, W&B Inference, Novita AI, Fireworks AI and Replicate.

What is Kimi K2.5's context window?

262K tokens, with up to 262K tokens in one reply.

Is Kimi K2.5 being retired?

Azure AI Foundry retires it on 26 Jan 2027. 9 hosts keep serving it.

Is there a newer version of Kimi K2.5?

Yes. Kimi K3.1 is the newest model in this line.

What are cheaper alternatives to Kimi K2.5?

Models that cost less and keep most of what it can do: Gemini 3.5 Flash Lite on Google AI Studio (29% less), GPT-5.4 Nano on OpenAI (61% less) and Kimi K2.7 Code on CLF AI Gateway (43% less). Kimi K2.5 itself costs 25% less on DeepInfra. Savings are on a mix of 3 input tokens to 1 output token.

When was Kimi K2.5 released?

Moonshot AI released it on 27 Jan 2026.

What is Kimi K2.5's API model id?

Use kimi-k2.5 with Moonshot AI. Other hosts use their own ids: Kimi-K2.5 on DeepInfra, moonshotai.kimi-k2.5 on Amazon Bedrock, Kimi-K2.5 on Baseten, Kimi-K2.5 on GMI Cloud.

Sources

Raw catalog entry for kimi-k2.5 on Moonshot AI
{
  "cache_read_input_token_cost": 1e-7,
  "deprecation_date": "2026-08-31",
  "input_cost_per_token": 6e-7,
  "litellm_provider": "moonshot",
  "max_input_tokens": 262144,
  "max_output_tokens": 262144,
  "mode": "chat",
  "output_cost_per_token": 0.000003,
  "source": "https://platform.moonshot.ai/docs/guide/kimi-k2-5-quickstart",
  "supports_function_calling": true,
  "supports_reasoning": true,
  "supports_response_schema": true,
  "supports_tool_choice": true,
  "supports_video_input": true,
  "supports_vision": true
}