Models/Moonshot/Kimi K2 0905

Kimi K2 0905

Kimi K2 0905 is the September update of [Kimi K2 0711](moonshotai/kimi-k2). It is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32...

01

Identity

262,144Context in
100,352Context out
2025-09-04 (OpenRouter created 1757021147)Released
OpenWeights
Canonical name
Kimi K2 0905
Aliases
moonshotai/kimi-k2-0905, moonshotai/Kimi-K2-Instruct-0905
Continuum hosted id
Not listed as a Continuum hosted id
OpenRouter slug
moonshotai/kimi-k2-0905
Hugging Face
moonshotai/Kimi-K2-Instruct-0905
Modalities
text
Open weights Not on Continuum host list Open

Context window on the 2026-08-19 OpenRouter row: 262,144 tokens.

Max completion tokens on that row: 100,352.

Input / output on that row: $0.60 / $2.50 per 1M tokens.

Architecture fields: text->text · Other.

Hugging Face id on that row: moonshotai/Kimi-K2-Instruct-0905.

02

Should I use this for coding agents

Kimi K2 0905 is the September update of [Kimi K2 0711](moonshotai/kimi-k2).

When you want the 2025-09-04 catalog SKU, not a later rename.

DeepSWE

No official DeepSWE mini-swe-agent row for this identity on the public board we fetched.

omitted, not invented as of 2026-08-13 source
Artificial Analysispublished integer
24
bench
Artificial Analysis
version
Intelligence Index
harness
published integer
metric
Intelligence Index
value
24
independent as of 2026-08-19 source
Vals Index

We did not open a vals.ai card for this identity. Chip omitted.

omitted, not invented as of 2026-08-19 source

Ranked view: Best coding models: the independent leaderboard puts this row and every other card that carries an independent coding score on one board, with the confidence intervals left visible.

Named benches, printed only

Copied from AA or vals HTML we opened. 0.0% placeholder bars are omitted. Do not average these into the hero tiles, and do not invent CI, $/task, or steps for Artificial Analysis.

BenchPrintedNoteSourceURLAs of
SciCode 30.7% Printed on the AA Intelligence Evaluations grid for Kimi K2 0905. Not a DeepSWE chip. Artificial Analysis Kimi K2 0905 source 2026-08-19
Humanity's Last Exam 6.4% Printed on the AA Intelligence Evaluations grid for Kimi K2 0905. Not a DeepSWE chip. Artificial Analysis Kimi K2 0905 source 2026-08-19
GPQA Diamond 76.7% Printed on the AA Intelligence Evaluations grid for Kimi K2 0905. Not a DeepSWE chip. Artificial Analysis Kimi K2 0905 source 2026-08-19
AA-Omniscience Accuracy 25.4% Printed on the AA Intelligence Evaluations grid for Kimi K2 0905. Not a DeepSWE chip. Artificial Analysis Kimi K2 0905 source 2026-08-19
AA-LCR 53.7% Printed on the AA Intelligence Evaluations grid for Kimi K2 0905. Not a DeepSWE chip. Artificial Analysis Kimi K2 0905 source 2026-08-19
03

When not to use it

Trust this before you buy

  • When you need a Continuum-hosted SKU from this lab, use the hosted card on this hub instead of this catalog row.
  • $0.60 in / $2.50 out per 1M on the 2026-08-19 OpenRouter row. Context 262,144 in / 100,352 out.
  • Need a denser published sibling on this hub? Kimi K3 is the card already on the cluster.
Get Plus See plans

Plus is $25/mo with $25 weekly hosted usage. The Mac app stays free with your own keys. Get Plus is not Download for Mac.

04

Artifact and Hugging Face downloads

Artifact

SPDX / license
other
Total params
HF safetensors.total: 1,026,470,735,448 stored tensors
Native precision
HF tensors F32 + BF16 + F8_E4M3
Repo size
HF API usedStorage 1,029,258,547,073 bytes
HF created
2025-09-03T03:34:36.000Z
Official HF repo
moonshotai/Kimi-K2-Instruct-0905
HF downloads
42,496
HF likes
786

Official weights id on the 2026-08-19 OpenRouter row: moonshotai/Kimi-K2-Instruct-0905. Community GGUF is a quant, not a second Continuum card.

Official weights

Use this model on Hugging Face Files on Hugging Face

This is the model repo, not Download for Mac. Downloads and likes from the Hugging Face API on 2026-08-19.

Hub file tree

Safetensor shards
62
Shard bytes
1,029,190,989,504 bytes
Other file bytes
17,089,021 bytes
Tree file count
80
Hub usedStorage
1,029,258,547,073 bytes

usedStorage and the sum of .safetensors sizes can differ (LFS pointers, non-shard files, duplicate copies). Cited separately. Source: recursive tree API, 2026-08-19.

PathBytes
model-00001-of-000062.safetensors497,640,784
model-00002-of-000062.safetensors17,066,593,248
model-00003-of-000062.safetensors17,066,593,248
model-00004-of-000062.safetensors17,066,593,248
model-00005-of-000062.safetensors17,066,593,248
model-00006-of-000062.safetensors17,066,593,248
model-00007-of-000062.safetensors17,066,593,248
model-00008-of-000062.safetensors17,066,593,248
model-00009-of-000062.safetensors17,066,593,248
model-00010-of-000062.safetensors17,066,593,248
model-00011-of-000062.safetensors17,066,595,576
model-00012-of-000062.safetensors17,066,595,576
model-00013-of-000062.safetensors17,066,595,576
model-00014-of-000062.safetensors17,066,595,576
model-00015-of-000062.safetensors17,066,595,576
model-00016-of-000062.safetensors17,066,595,576
model-00017-of-000062.safetensors17,066,595,576
model-00018-of-000062.safetensors17,066,595,576
model-00019-of-000062.safetensors17,066,595,576
model-00020-of-000062.safetensors17,066,595,576
model-00021-of-000062.safetensors17,066,595,576
model-00022-of-000062.safetensors17,066,595,576
model-00023-of-000062.safetensors17,066,595,576
model-00024-of-000062.safetensors17,066,595,576
model-00025-of-000062.safetensors17,066,595,576
model-00026-of-000062.safetensors17,066,595,576
model-00027-of-000062.safetensors17,066,595,576
model-00028-of-000062.safetensors17,066,595,576
model-00029-of-000062.safetensors17,066,595,576
model-00030-of-000062.safetensors17,066,595,576
model-00031-of-000062.safetensors17,066,595,576
model-00032-of-000062.safetensors17,066,595,576
model-00033-of-000062.safetensors17,066,595,576
model-00034-of-000062.safetensors17,066,595,576
model-00035-of-000062.safetensors17,066,595,576
model-00036-of-000062.safetensors17,066,595,576
model-00037-of-000062.safetensors17,066,595,576
model-00038-of-000062.safetensors17,066,595,576
model-00039-of-000062.safetensors17,066,595,576
model-00040-of-000062.safetensors17,066,595,576
model-00041-of-000062.safetensors17,066,595,576
model-00042-of-000062.safetensors17,066,595,576
model-00043-of-000062.safetensors17,066,595,576
model-00044-of-000062.safetensors17,066,595,576
model-00045-of-000062.safetensors17,066,595,576
model-00046-of-000062.safetensors17,066,595,576
model-00047-of-000062.safetensors17,066,595,576
model-00048-of-000062.safetensors17,066,595,576
model-00049-of-000062.safetensors17,066,595,576
model-00050-of-000062.safetensors17,066,595,576
model-00051-of-000062.safetensors17,066,595,576
model-00052-of-000062.safetensors17,066,595,576
model-00053-of-000062.safetensors17,066,595,576
model-00054-of-000062.safetensors17,066,595,576
model-00055-of-000062.safetensors17,066,595,576
model-00056-of-000062.safetensors17,066,595,576
model-00057-of-000062.safetensors17,066,595,576
model-00058-of-000062.safetensors17,066,595,576
model-00059-of-000062.safetensors17,066,595,576
model-00060-of-000062.safetensors17,066,595,576
model-00061-of-000062.safetensors17,066,595,576
model-00062-of-000062.safetensors4,697,635,112
model.safetensors.index.json13,525,772
tiktoken.model2,795,286
figures/banner.png291,736
figures/Base-Evaluation.png245,449
figures/kimi-logo.png87,988
modeling_deepseek.py75,769
tokenization_kimi.py12,531
docs/tool_call_guidance.md11,343
configuration_deepseek.py10,652
README.md9,894
docs/deploy_guidance.md8,896
tokenizer_config.json3,695
config.json3,174
chat_template.jinja2,023
THIRD_PARTY_NOTICES.md1,664
.gitattributes1,634
LICENSE1,463
generation_config.json52

All 80 files the tree API returned are listed: every one of the 62 safetensor shards plus 18 other files. Files tab: https://huggingface.co/moonshotai/Kimi-K2-Instruct-0905/tree/main.

05

Continuum serving

OpenRouter slug moonshotai/kimi-k2-0905 on the 2026-08-19 catalog.

First party: https://platform.moonshot.ai/docs/overview.

Not on the Continuum host list fetched 2026-08-19.

Continuum hosted id not on the host list
Kimi K2 0905 first partyOpenRouter listing

We list first-party and OpenRouter, plus Continuum only when the live public allowlist named the id. This is not a 15-host routing table.

06

Economics

OpenRouter 2026-08-19: $0.60 in / $2.50 out per 1M tokens. Context 262,144 in / 100,352 out.

No separate internal-reasoning price on that row.

Price this model Opens the pricing calculator preloaded with Kimi K2 0905.
07

Provenance

ClaimSourceAs of
OpenRouter id moonshotai/kimi-k2-0905; context 262144; created 1757021147; $0.60 / $2.50 per 1MOpenRouter /api/v1/models2026-08-19
AA Intelligence Index 24; 5 printed Index benchesArtificial Analysis model page2026-08-19
OpenRouter id moonshotai/kimi-k2-0905, context 262,144OpenRouter /api/v1/models2026-08-19
HF downloads 42,496Hugging Face API moonshotai/Kimi-K2-Instruct-09052026-08-19
HF safetensors.total: 1,026,470,735,448 stored tensors; HF API usedStorage 1,029,258,547,073 bytes; created 2025-09-03T03:34:36.000Z; HF tensors F32 + BF16 + F8_E4M3Hugging Face API moonshotai/Kimi-K2-Instruct-09052026-08-19
62 safetensor shards; shard bytes 1,029,190,989,504; tree files 80Hugging Face tree API moonshotai/Kimi-K2-Instruct-09052026-08-19
08

Compare, FAQ, and Get Plus

No other card in this cluster shares a published DeepSWE official row we can put next to this one. We will not compare on vendor-blog numbers.

Continue through Moonshot's model family with Kimi K2 0711.

FAQ

Why does the lab blog disagree with DeepSWE or Scale?

Lab posts pick a harness, an effort, a split, and sometimes a private eval set. DeepSWE publishes the official mini-swe-agent row with cost, tokens, and steps. Scale SWE-bench Pro only counts when the public shared-harness board has a row we can fetch. A higher lab number is usually a different test, not a better one.

OpenAI-compatible call

No Continuum hosted id is verified for this model, so this page does not invent a curl target. Use the first-party API or the OpenRouter slug moonshotai/kimi-k2-0905.

Get Plus See plans

Plus is $25/mo with $25 weekly hosted usage. The Mac app stays free with your own keys. Get Plus is not Download for Mac.