kimi-k2

completions

On:deepinfraparasail

kimi-k2

completions

Published by: moonshotaiProviders:

deepinfra parasail

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.

Provider	Input	Output
deepinfra	$0.5 / 1M tokens	$2 / 1M tokens
parasail	$0.99 / 1M tokens	$2.99 / 1M tokens

Released

Jan 1, 2025

Knowledge

Jul 5, 2024

Context

131K

Input

$0.5 / 1M tokens

Output

$2 / 1M tokens

Capabilities: tools

Accepts: text

Returns: text

Providers	Context	Input Price	Output Price	Input Formats	Output Formats	License
deepinfra	131K tokens	$0.5 / 1M tokens	$2 / 1M tokens	text	text	Modified MIT License
parasail	131072 tokens	$0.99 / 1M tokens	$2.99 / 1M tokens	text	text	Modified MIT License

Released Jan 1, 2025Knowledge Cutoff: Jul 5, 2024

Access kimi-k2 through LangDB AI Gateway

Recommended

Integrate with moonshotai's kimi-k2 and 250+ other models through a unified API. Monitor usage, control costs, and enhance security.

Unified API

Cost Optimization

Enterprise Security

Get Started Now

Free tier available • No credit card required

Instant Setup

99.9% Uptime

10,000+Monthly Requests

Available from 2 providers

Provider:

Benchmark Results for kimi-k2

Category Performance Scores:

Academia: Score 63.33 (Top 11% - Rank #38)
Finance: Score 53.85 (Top 21% - Rank #73)
Marketing: Score 57.60 (Top 16% - Rank #56)
Maths: Score 57.30 (Top 27% - Rank #94)
Science: Score 70.79 (Top 18% - Rank #63)
Writing: Score 56.12 (Top 24% - Rank #83)

Overall Performance: 59.83111111111111 average score across all categories

Detailed Benchmark Scores:

Benchmark	Score	Percentile	Domain
AAII	50.40	Top 9%	General
AA Math Index	57.30	Top 27%	Mathematics
GPQA	76.25	Top 15%	STEM (Physics, Chemistry, Biology)
HLE	6.30	Top 20%	General Knowledge
HumanEval	94.50	Top 0%	Programming
LiveCodeBench	61.00	Top 19%	Programming
MMLU	90.20	Top 4%	General Knowledge
MMLU-Pro	82.20	Top 16%	General Knowledge
SciCode	30.70	Top 38%	Scientific

MMLU Score: 90.20 - Massive Multitask Language Understanding benchmark

HumanEval Score: 94.50 - Coding capability benchmark

GPQA Score: 76.25 - Graduate-level reasoning benchmark

Model Comparison:

Provider: deepinfra

Model Type: completions

Context Size: 131000 tokens

Comparing against 345 models in the database

Performance Analytics for kimi-k2

Usage Statistics (Last 1 Days):

Total Requests: 9 API calls
Average TPS: 1415.63 tokens per second
Average Response Time: 4978.50ms
Average Time to First Token: 423.80ms
Total Cost: $0.04
Average Request Cost: $0.0046

Daily Performance Breakdown:

Date	Requests	TPS	Response Time	TTFT	Cost
9/20/2025	9	1415.63	4978.50ms	423.80ms	$0.04

Performance Summary:

Model: kimi-k2 by deepinfra

Monitoring Period: 9/13/2025 to 9/20/2025

Average Daily Requests: 9

Peak Daily Requests: 9

Statistics

View Full Statistics

Request Volume

Daily API requests

Performance (TPS)

Tokens per second

1415.63 tokens/s

Category Scores

Benchmark Tests

View Other Benchmarks

AAII

50.4

General

AA Math Index

57.3

Mathematics

GPQA

76.3

STEM (Physics, Chemistry, Biology)

HLE

6.3

General Knowledge

HumanEval

94.5

Programming

LiveCodeBench

61.0

Programming

MMLU

90.2

General Knowledge

MMLU-Pro

82.2

General Knowledge

SciCode

30.7

Scientific

Metric	AAII	AA Math Index	GPQA	HLE	HumanEval	LiveCodeBench	MMLU	MMLU-Pro	SciCode
Score	50.4	57.3	76.3	6.3	94.5	61.0	90.2	82.2	30.7

Compare with Similar Models

gemini-2.5-pro-preview

grok-4

Code Examples

Integration samples and API usage

Code Samples for kimi-k2

Python SDK Example:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.us-east-1.langdb.ai/projects/<your_project_id>",
    api_key="<your_api_key>"
)

response = client.chat.completions.create(
    model="kimi-k2",
    messages=[
        {"role": "user", "content": "Hello, how are you?"}
    ]
)

print(response.choices[0].message.content)

TypeScript SDK Example:

import OpenAI from 'openai';

const client = new OpenAI({
    baseURL: "https://api.us-east-1.langdb.ai/projects/<your_project_id>",
    apiKey: "<your_api_key>"
});

const response = await client.chat.completions.create({
    model: "kimi-k2",
    messages: [
        { role: "user", content: "Hello, how are you?" }
    ]
});

console.log(response.choices[0].message.content);

cURL Example:

curl -X POST "https://api.us-east-1.langdb.ai/projects/<your_project_id>/v1/chat/completions" \
  -H "Authorization: Bearer <your_api_key>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k2",
    "messages": [
        {"role": "user", "content": "Hello, how are you?"}
    ]
  }'

Model: kimi-k2

Provider: deepinfra

API Endpoint: $https://api.us-east-1.langdb.ai

Create API Key

Related Models

Similar models from deepinfra

kimi-k2

kimi-k2

Access kimi-k2 through LangDB AI Gateway

Statistics

Category Scores

Benchmark Tests

Compare with Similar Models

Code Examples

Related Models

deepseek-chat-v3-0324

deepseek-chat-v3.1

deepseek-prover-v2

DeepSeek-R1

deepseek-r1-0528

DeepSeek-R1-Distill-Llama-70B

kimi-k2 by deepinfra - AI Model Details, Pricing, and Performance Metrics

kimi-k2

kimi-k2

Access kimi-k2 through LangDB AI Gateway

Statistics

Category Scores

Benchmark Tests

Compare with Similar Models

Code Examples

Related Models

deepseek-chat-v3-0324

deepseek-chat-v3.1

deepseek-prover-v2

DeepSeek-R1

deepseek-r1-0528

DeepSeek-R1-Distill-Llama-70B