impulse

Kimi K2 Thinking

Open weights

Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning.

API id Kimi-K2-ThinkingReleased 6 Nov 2025 (Nov 2025)
Amazon Bedrock retires it on 30 Mar 2027 (in 132 days). 8 hosts keep serving it.
Input / 1M
$0.80
at GMI Cloud
Output / 1M
$1.20
at GMI Cloud
Context window
262K
tokens
Max output
16K
tokens per reply
Hosts
9
17 listings
Released
Nov 2025
6 Nov 2025

Pricing

List prices at GMI Cloud, in US dollars per million tokens.

TierInput / 1MOutput / 1M
Standard$0.80$1.20

What it costs a month

At GMI Cloud list prices, 30 days.

  • Chat assistant5,000 requests a day, 2K in / 0.5K out$330
  • RAG over documents2,000 requests a day, 8K in / 0.6K out$427
  • Coding agent1,000 requests a day, 40K in / 2K out$1,032
  • Overnight batch20,000 requests a day, 4K in / 0.4K out$2,208
Use your own numbers

Where to run it

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

HostIn / out per 1M
GMI CloudInference
Kimi-K2-Thinking
$0.80 / $1.20
kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
kimi-k2-thinking-maas
$0.60 / $2.50
+19% vs headline
moonshot.kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
Retires 30 Mar 2027
BasetenInference
Kimi-K2-Thinking
$0.60 / $2.50
+19% vs headline
Fireworks AIInference
kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
Novita AIInference
kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
ReplicateInference
kimi-k2-thinking
$0.66 / $2.70
+30% vs headline
CrusoeInference
Kimi-K2-Thinking
$2.50 / $2.50
+178% vs headline
Show 8 more listings: regions, snapshots and variants
Amazon Bedrock
Alias
moonshotai.kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
Retires 30 Mar 2027
us-east-1
moonshotai.kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
Retires 30 Mar 2027
us-east-2
moonshotai.kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
Retires 30 Mar 2027
us-west-2
moonshotai.kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
Retires 30 Mar 2027
Mantle
moonshotai.kimi-k2-thinking
$0.60 / $2.50
+19% vs headline
Retires 30 Mar 2027
ap-south-1
moonshotai.kimi-k2-thinking
$0.71 / $2.94
+41% vs headline
Retires 30 Mar 2027
ap-northeast-1
moonshotai.kimi-k2-thinking
$0.73 / $3.03
+45% vs headline
Retires 30 Mar 2027
sa-east-1
moonshotai.kimi-k2-thinking
$0.73 / $3.03
+45% vs headline
Retires 30 Mar 2027

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
Text
Produces
Text
Open weights
License: Modified MIT License
Runs in about 2463.4 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

No host sells it for less than GMI Cloud.

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
    6% less
  • GPT-5.4 Nano
    OpenAI · on OpenAI
    $0.20 / $1.25
    Has every key capability
    49% less
  • Claude Haiku 5.5
    Anthropic · on Anthropic
    $0.10 / $0.50
    Has every key capability
    78% less
  • GPT-6 Luna
    OpenAI · on OpenAI
    $0.10 / $0.50
    Has every key capability
    78% less

Open-weight picks

  • Motif 3
    Independent · on ai&
    $0.50 / $2.00
    Has every key capability
    3% less
  • GLM 5.3
    Zhipu AI · on CLF AI Gateway
    $0.56 / $1.76
    Has every key capability
    4% less
  • Nemotron 3 Ultra 550B A55B
    NVIDIA · on Perplexity
    $0.25 / $2.50
    Has every key capability
    10% less
  • DeepSeek V4 Pro
    DeepSeek · on CLF AI Gateway
    $0.528 / $1.58
    Has every key capability
    12% less

Call it

Using GMI Cloud's API.

Python
from litellm import completion

response = completion(
    model="gmi/Kimi-K2-Thinking",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Quick answers

How much does Kimi K2 Thinking cost?

On GMI Cloud, Kimi K2 Thinking costs $0.80 per million input tokens and $1.20 per million output tokens.

Which APIs offer Kimi K2 Thinking?

Kimi K2 Thinking runs on 9 APIs: GMI Cloud, Azure AI Foundry, Vertex AI, Amazon Bedrock, Baseten, Fireworks AI, Novita AI, Replicate and Crusoe.

What is Kimi K2 Thinking's context window?

262K tokens, with up to 16K tokens in one reply.

Is Kimi K2 Thinking being retired?

Amazon Bedrock retires it on 30 Mar 2027. 8 hosts keep serving it.

What are cheaper alternatives to Kimi K2 Thinking?

Models that cost less and keep most of what it can do: Gemini 3.5 Flash Lite on Google AI Studio (6% less), GPT-5.4 Nano on OpenAI (49% less) and Motif 3 on ai& (3% less). Savings are on a mix of 3 input tokens to 1 output token.

When was Kimi K2 Thinking released?

Moonshot AI released it on 6 Nov 2025.

What is Kimi K2 Thinking's API model id?

Use Kimi-K2-Thinking with GMI Cloud. Other hosts use their own ids: kimi-k2-thinking on Azure AI Foundry, kimi-k2-thinking-maas on Vertex AI, moonshot.kimi-k2-thinking on Amazon Bedrock, kimi-k2-thinking on Fireworks AI.

Sources

Raw catalog entry for Kimi-K2-Thinking on GMI Cloud
{
  "input_cost_per_token": 8e-7,
  "litellm_provider": "gmi",
  "max_input_tokens": 262144,
  "max_output_tokens": 16384,
  "mode": "chat",
  "output_cost_per_token": 0.0000012
}