Singularity Logs

THE MODEL, IN DETAIL / moonshotai

Kimi-K2.5

moonshotai/kimi-k2.5Open weightsReasoning

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...

OpenRouter
Source-linked model intelligenceCatalog added 27 Jan 2026Specifications observed 17 Oct 2026

GPQA Diamond

87.6%

Source-reported

Context window

262.1K

235.9K max output tokens

Input price

$0.5

USD / million input tokens

Output price

$2.5

USD / million output tokens

Inference providers

5

5 provider routes observed

Measured output speed

—

No model-level measurement

01 / PERFORMANCE

A clearer view of capability.

Read the score, understand the context, and explore where this model sits among its peers.

Performance fingerprint

Individual published accuracy scores · not a composite index

Individual benchmark accuracy. Axes represent different tasks, not a composite index.AIME 2026GPQA DiamondHumanity’s Last…HMMT February 2…MMLU-ProAIME 2026: 95.8GPQA Diamond: 87.6Humanity’s Last Exam: 50.2HMMT February 2026: 87.1MMLU-Pro: 87.1Each axis: 0–100% · Individual evaluations

Performance meets price

GPQA Diamond · same published source and version

Higher score · lower cost70$075$0.287580$0.57585$0.862590$1.15Qwen3.5-397B-A17B GPQA Diamond: 88.4 Input: $0.5/M tokensKimi-K2.5 GPQA Diamond: 87.6 Input: $0.5/M tokensKimi-K2.5Qwen3.5-122B-A10B GPQA Diamond: 86.6 Input: $0.26/M tokensGLM-5 GPQA Diamond: 86 Input: $1/M tokensGLM-4.7 GPQA Diamond: 85.7 Input: $0.54/M tokensQwen3.5-27B GPQA Diamond: 85.5 Input: $0.26/M tokensMiniMax-M2.5 GPQA Diamond: 85.2 Input: $0.27/M tokensKimi-K2-Thinking GPQA Diamond: 84.5 Input: $0.6/M tokensQwen3.5-35B-A3B GPQA Diamond: 84.2 Input: $0.15/M tokensStep-3.5-Flash GPQA Diamond: 83.5 Input: $0.1/M tokensDeepSeek-V3.2 GPQA Diamond: 82.4 Input: $0.28/M tokensInput price · USD / million tokens
Qwenmoonshotaizai-orgMiniMaxAIstepfun-aideepseek-ai

In good company

GPQA Diamond · OpenEvals / Hugging Face

88.4
Qwen
87.6
This model
86.6
Qwen
86
zai-org
85.7
zai-org
85.5
Qwen
85.2
MiniMaxAI
84.5
moonshotai
84.2
Qwen
83.8
Nanbeige

Rank #2 in this published cohort. 27 models in the source cohort. Showing selected nearby models. Evaluation settings can differ.

Evidence, by task.

Coverage is not a capability score.

5published results
1 evaluation source
Reasoning3 evaluations
CodingAwaiting evidence
Mathematics2 evaluations
VisionAwaiting evidence
AgenticAwaiting evidence
Long contextAwaiting evidence

Documented features, task evaluations, and preference rankings describe different things. Missing coverage does not imply a model cannot perform a task.

Explore our methodology

02 / BENCHMARK EXPLORER

The numbers behind the model.

Original results, exact versions, and a direct path to the evidence behind every score.

5 published results
020406080100AIME 2026: 95.8% · Published snapshot 2026-03-16 · OpenEvals / Hugging Face95.8%AIME 2026Higher is betterGPQA Diamond: 87.6% · Published snapshot 2026-03-16 · OpenEvals / Hugging Face87.6%GPQA DiamondHigher is betterHumanity’s Last Exam: 50.2% · Published snapshot 2026-03-16 · OpenEvals / Hugging Face50.2%Humanity’s Last …Higher is betterHMMT February 2026: 87.1% · Published snapshot 2026-03-16 · OpenEvals / Hugging Face87.1%HMMT February 2026Higher is betterMMLU-Pro: 87.1% · Published snapshot 2026-03-16 · OpenEvals / Hugging Face87.1%MMLU-ProHigher is better

5 percentage-based evaluations. Each benchmark measures a different task; compare results within the same evaluation and protocol.

Published benchmark results for Kimi-K2.5
BenchmarkScoreVersion & metricSourcePublishedEvidence
Published snapshot 2026-03-16Accuracy · Higher is betterOpenEvals / Hugging FaceSource-reported15 Mar 2026
Published snapshot 2026-03-16Accuracy · Higher is betterOpenEvals / Hugging FaceSource-reported15 Mar 2026
Published snapshot 2026-03-16Accuracy · Higher is betterOpenEvals / Hugging FaceSource-reported15 Mar 2026
Published snapshot 2026-03-16Accuracy · Higher is betterOpenEvals / Hugging FaceSource-reported15 Mar 2026
Published snapshot 2026-03-16Accuracy · Higher is betterOpenEvals / Hugging FaceSource-reported15 Mar 2026

Artificial Analysis indices

Published in the OpenRouter catalog · source-defined index values

Coding Index46.8
Intelligence Index23.5

Catalog observation: 17 Oct 2026. Index versions and evaluation settings are not supplied in this feed. These are not Singularity composite scores.

Artificial Analysis

Design Arena

Source-reported Elo across individual design categories

Androidnative1,133
agents arena · Rank #26 · 57.8% win rate
Fullstack1,114
agents arena · Rank #28 · 54.5% win rate
Godotgamedev1,219
agents arena · Rank #15 · 59.8% win rate
Mobileapps1,145
agents arena · Rank #28 · 54.7% win rate
Webapps1,129
agents arena · Rank #29 · 51.6% win rate
3d1,206
models arena · Rank #47 · 53.1% win rate
Asciiart1,168
models arena · Rank #33 · 46.4% win rate
Codecategories1,238
models arena · Rank #45 · 54.1% win rate
Dataviz1,220
models arena · Rank #51 · 51.3% win rate
Gamedev1,208
models arena · Rank #50 · 53.4% win rate
Svg1,149
models arena · Rank #44 · 48.4% win rate
Uicomponent1,236
models arena · Rank #45 · 53.6% win rate
Website1,250
models arena · Rank #41 · 55.2% win rate

Catalog observation: 17 Oct 2026. Category ratings have separate comparison pools.

03 / COST & VALUE

Know what your workload costs.

Explore token prices, see where the spend goes, and model a workload of your own.

Input$0.5per million tokens
Output$2.5per million tokens
Cache read$0.15per million tokens
Cache write—per million tokens

Workload cost calculator

Change the numbers. See the trade-offs.

Token estimate only. Excludes cache writes, tools, and other provider fees. Actual token usage and provider limits vary.

Your estimated spend

1,000 requests · 11M tokens

$7.5estimated total
Uncached input$5
Cached input$0
Output$2.5
Per request$0.0075
Without cache discount$7.5
Estimated cache savings$0

04 / INFERENCE NETWORK

One model. Different ways to run it.

Compare inference routes by price, context, quantization, and reported availability.

5 providers · 5 routes
Provider routes for Kimi-K2.5, observed 17 Oct 2026
Provider / routeInput / MOutput / MCache / MContextOutput limitSpeedLatencyAvailability · 1dRates
AAtlasCloudINT4$0.49$2.5$0.2262.1K235.9K——98.23%
SSiliconFlowINT4$0.5$2.5$0.15262.1K235.9K——100%
VVeniceQuantization not disclosed$0.532$3.325$0.209256K65.5K——100%
NNovitaQuantization not disclosed$0.57$2.85$0.095262.1K235.9K——99.32%
AAmazon BedrockQuantization not disclosed$0.6$3—262.1K131.1K——100%

Snapshot observed 17 Oct 2026. Speed and latency are the reported 30-minute median when available. This is a dated snapshot, not live monitoring.

Provider price comparison

USD per million tokens · input and output rates

InputOutput
AtlasCloudint4
Input
$0.49
Output
$2.5
SiliconFlowint4
Input
$0.5
Output
$2.5
Venice
Input
$0.532
Output
$3.325
Novita
Input
$0.57
Output
$2.85
Amazon Bedrock
Input
$0.6
Output
$3

Reported availability

Provider availability over the selected observation window

AtlasCloud · INT498.23%
SiliconFlow · INT4100%
Venice100%
Novita99.32%
Amazon Bedrock100%

Output speed

Reported 30-minute p50 · tokens per second

No provider throughput measurements were returned in this snapshot.

Response latency

Reported 30-minute p50 · seconds

No provider latency measurements were returned in this snapshot.

Context by provider

Actual route limits can differ from the catalog maximum

AtlasCloud · INT4262.1K
SiliconFlow · INT4262.1K
Venice256K
Novita262.1K
Amazon Bedrock262.1K

05 / CAPABILITIES & MODALITIES

What goes in. What comes out.

A map of documented interfaces and API features, separate from benchmark performance.

INPUT MODALITIES

TextImage
Kimi-K2.5Reasoning supported

OUTPUT MODALITIES

Text

Feature support

Listed catalog API parameters and documented pricing features

ReasoningSupported
Tool callingSupported
Tool choiceSupported
Structured outputsSupported
Response formatSupported
Reasoning detailsSupported
Seed controlSupported
Stop sequencesSupported
Prompt cachingSupported
TemperatureSupported

Room to work

Context capacity and completion limits

262.1Ktotal context tokens
Context window
235.9Kmax output

Output tokens share the total context budget. The maximum completion length is a separate limit, and individual provider routes may have smaller limits.

Catalog context
262,144 tokens
Maximum output
235,929 tokens
Top-provider context
262,144 tokens

Features vary by route

Check the provider before depending on a specific API feature.

API feature support by provider
Provider routeReasoningToolsStructured outputResponse formatImplicit caching
AtlasCloud · INT4SupportedSupportedSupportedSupportedSupported
SiliconFlow · INT4SupportedSupportedSupportedSupportedSupported
VeniceSupportedSupportedSupportedSupportedSupported
NovitaSupportedSupportedSupportedSupportedSupported
Amazon BedrockSupportedSupportedNot listedNot listedNot listed

06 / UNDER THE HOOD

The complete specification sheet.

Exact identities, model architecture, limits, and the defaults that shape a request.

Identity & architecture

Developer
moonshotai
Exact identity
moonshotai/Kimi-K2.5
Subject
Model
Access
Open weights
License
Not disclosed
Parameters
1,058.6 billion
Tokenizer
Other
Instruction format
Not disclosed

Reasoning & controls

Reasoning
Supported
Mandatory reasoning
No
Reasoning by default
Enabled
Published configuration
Not disclosed
Supported reasoning efforts
Not documented
Default reasoning effort
Not disclosed
API parameters
19
Provider moderation
No
Quantization variants
int4

Dates & limits

Catalog added
27 Jan 2026
Knowledge cutoff
Not disclosed
Catalog expiration
Not documented
Specifications observed
17 Oct 2026
Providers observed
17 Oct 2026
Request limits
Provider-specific / not supplied
Canonical catalog ID
moonshotai/kimi-k2.5-0127

API parameters & defaults

Availability is catalog-wide. Defaults can vary across providers.

frequency_penaltySupported · default not supplied
include_reasoningSupported · default not supplied
logit_biasSupported · default not supplied
logprobsSupported · default not supplied
max_tokensSupported · default not supplied
min_pSupported · default not supplied
presence_penaltySupported · default not supplied
reasoningSupported · default not supplied
repetition_penaltySupported · default not supplied
response_formatSupported · default not supplied
seedSupported · default not supplied
stopSupported · default not supplied
structured_outputsSupported · default not supplied
temperatureSupported · default not supplied
tool_choiceSupported · default not supplied
toolsSupported · default not supplied
top_kSupported · default not supplied
top_logprobsSupported · default not supplied
top_pSupported · default not supplied

07 / BUILD WITH THIS MODEL

From exploration to your first request.

Use the documented catalog identifier and keep your API key in an environment variable.

Make an API request

const response = await fetch(
  "https://openrouter.ai/api/v1/chat/completions",
  {
    method: "POST",
    headers: {
      Authorization: `Bearer ${process.env.OPENROUTER_API_KEY}`,
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      "model": "moonshotai/kimi-k2.5",
      "messages": [
        {
          "role": "user",
          "content": "Explain the trade-offs in this design."
        }
      ]
    }),
  },
);
if (!response.ok) throw new Error(`HTTP ${response.status}`);
const data = await response.json();
console.log(data.choices[0].message.content);

08 / PROVENANCE & HISTORY

Every number has a story.

Follow the source, check the observation date, and distinguish published results from model specifications.

The observation trail

Recorded events, in chronological order

Added to the model catalog

Catalog creation date, as recorded by OpenRouter.

AIME 2026 result published

95.8% · OpenEvals / Hugging Face

GPQA Diamond result published

87.6% · OpenEvals / Hugging Face

Humanity’s Last Exam result published

50.2% · OpenEvals / Hugging Face

HMMT February 2026 result published

87.1% · OpenEvals / Hugging Face

MMLU-Pro result published

87.1% · OpenEvals / Hugging Face

Model specifications observed

Catalog pricing, limits, and supported parameters captured.

Provider routes observed

5 inference routes in this snapshot.

Observation dates describe when data was captured. They do not represent newly run evaluations or continuous performance monitoring.

Published observation ledger

Historical evidence, with source and version preserved.

Published observation history for Kimi-K2.5
ObservedBenchmarkScoreVersionSourceVerification
13 Oct 2026HMMT February 2026Published snapshot 2026-03-16OpenEvals / Hugging FaceSource-reported
13 Oct 2026AIME 2026Published snapshot 2026-03-16OpenEvals / Hugging FaceSource-reported
13 Oct 2026Humanity’s Last ExamPublished snapshot 2026-03-16OpenEvals / Hugging FaceSource-reported
13 Oct 2026GPQA DiamondPublished snapshot 2026-03-16OpenEvals / Hugging FaceSource-reported
13 Oct 2026MMLU-ProPublished snapshot 2026-03-16OpenEvals / Hugging FaceSource-reported

09 / A FEW MORE DETAILS

Questions worth asking.

Quick answers, drawn from this model’s connected specifications.

How much does Kimi-K2.5 cost?

Catalog starting rates are $0.5 per million input tokens and $2.5 per million output tokens. Cache-read pricing is $0.15 per million tokens. Compare provider routes and service tiers above for their specific prices.

How much context can it handle?

262,144 tokens in the catalog context window. The catalog lists a maximum of 235,929 completion tokens. Provider limits can differ.

Does it support tools and structured output?

Tool calling: listed as supported. Structured outputs: listed as supported. See the provider feature matrix for route-specific differences.

How should I interpret these benchmark results?

Read the benchmark name, exact version, metric, source, and model configuration together. Different evaluation settings can change the result. Preference ratings and source-defined composite indices remain separate from task accuracy.

Are these prices and provider measurements live?

This page uses dated source snapshots. Catalog data was observed on 17 Oct 2026. Provider routes were observed on 17 Oct 2026. Missing measurements are not estimated or treated as zero.

KEEP EXPLORING

More from moonshotai.

Explore related identities from the same developer, or take them into a side-by-side comparison.

THE RIGHT MODEL STARTS WITH THE RIGHT CONTEXT.

Put the details side by side.

Explore the trade-offs in a direct model comparison.

Compare Kimi-K2.5
Source-backed evidence. Different tasks. A clearer picture.
AboutMethodologyBenchmark dashboard