Open-Source LLM July 17, 2026 ~24 min read Kimi K3 Moonshot AI

Kimi K3 Review
The 2.8-Trillion-Parameter Open-Source Model That Challenges Claude and GPT

July 16, 2026 launch · KDA architecture · 1M context · full benchmarks · API pricing · July 27 open weights

Kimi K3 Moonshot AI 2.8 trillion parameter open-source large language model

TL;DR: Moonshot AI just released Kimi K3 — the world's largest open-source AI model at 2.8 trillion parameters. It has a 1M token context window, native vision, beats Claude Fable 5 and GPT-5.6 Sol on several coding benchmarks, and costs $3/$15 per million tokens. Full weights drop July 27. Below: architecture, benchmarks, pricing, how to try it, and whether you should switch.

01

What Is Kimi K3?

Kimi K3 is a 2.8-trillion-parameter mixture-of-experts (MoE) model from Beijing-based Moonshot AI. It's the world's first open model in the 3T-class, surpassing the previous record holder DeepSeek V4 Pro (1.6T) by nearly 75%.

SpecDetail
Total Parameters2.8 trillion
ArchitectureKimi Delta Attention (KDA) + Attention Residuals + Stable LatentMoE
Active Experts16 out of 896 (sparse MoE)
Context Window1,048,576 tokens (1M)
Input ModalitiesText, image, video
API Model IDkimi-k3
Pricing$3 / $15 per 1M tokens (input/output)
Open WeightsJuly 27, 2026

The model is live on kimi.com, the Kimi mobile app, Kimi Code, and the Moonshot API. You can try it now for free with a Google account — no credit card needed.

Pain points this release targets

  1. 01

    Context truncation on long code tasks — 200K closed models force chunking on large repos

  2. 02

    Open-weight scale ceiling — prior max was ~1.6T, limiting frontier coding agents

  3. 03

    Theoretical vs practical long context — KV cache costs make advertised windows unusable in production

  4. 04

    Closed API dependency — no self-host fallback when pricing or policy shifts

02

Why This Release Matters

The last 18 months were rough for Moonshot AI. DeepSeek's rise eroded their market position. Kimi K3 is a striking comeback:

  • For 9 of the past 12 months, Kimi models held the largest open-source parameter record
  • ARR crossed $300M as of June 2026; 6th funding round at $31.5B pre-money valuation
  • API revenue is over 70% of total; overseas paid users up 400%
  • Timed for the eve of WAIC 2026 in Shanghai

This isn't a vanity project — it's a fast-growing business making a serious technical statement.

03

Architecture: Three Genuine Innovations

1. Kimi Delta Attention (KDA)

Full attention scales quadratically with context. At 1M tokens, KV cache memory becomes catastrophic. KDA alternates cheap linear-attention layers with full-attention layers in a 3:1 ratio:

  • Up to 75% less KV cache memory
  • Up to 6.3× faster decoding at 1M-token contexts
  • Matches or beats full-attention baselines on short, long, and RL tasks — no capability tradeoff

2. Attention Residuals (AttnRes)

Standard residuals dilute early-layer signals across depth. AttnRes enables selective retrieval across depth — pulling high-value representations from earlier layers. Result: ~25% higher training efficiency at under 2% extra compute.

3. Stable LatentMoE

Just 16 of 896 experts activate per forward pass (1.8% sparsity). Moonshot's fixes: Quantile Balancing, Per-Head Muon, SiTU, and Gated MLA. Together: ~2.5× better scaling efficiency vs Kimi K2.

04

Benchmark Results

BenchmarkKimi K3Claude Fable 5GPT-5.6 SolClaude Opus 4.8
DeepSWE67.570.073.059.0
Program Bench77.876.877.671.9
Terminal Bench 2.188.384.688.884.6
FrontierSWE81.286.671.366.7
SWE Marathon42.035.039.040.0
BrowseComp91.288.090.484.3
Automation Bench30.829.129.727.2
GPQA-Diamond93.592.694.191.0
MMMU-Pro (vision)81.681.283.078.9
OmniDocBench91.189.885.887.9

SWE Marathon — sustained long-horizon coding — is the headline: K3 leads at 42.0, a 7-point gap over Claude Fable 5. On Artificial Analysis Intelligence Index v4.1, K3 scores 57.1 (4th), behind Fable 5 (59.9) and GPT-5.6 Sol (58.9) — a 2.8-point gap from #1 to #4.

Caveat: Moonshot self-reported benchmarks. Different harnesses (Kimi Code, Codex, Claude Code). Treat as directionally useful, not definitive.

05

Pricing

ModelInput $/1MOutput $/1MCache-Hit InputContext
Kimi K3$3.00$15.00$0.301M tokens
Claude Sonnet 5 (standard)$3.00$15.00200K
Claude Opus 4.8$5.00$25.00200K
GPT-5.5$5.00$30.00400K
DeepSeek V4 Pro$1.74$3.48$0.145128K

K3 matches Claude Sonnet 5 standard pricing but gives 5× the context. Kimi Code workflows hit cache at 90%+ via Mooncake split-inference — effective average input cost can drop to roughly $0.55/M tokens in real usage. vs Opus 4.8: similar or better on several benchmarks at 60% input / 40% output cost.

06

How to Use Kimi K3 Right Now

  1. 01

    Just chatkimi.com, Google sign-up, max reasoning effort, no credit card

  2. 02

    API — key at platform.kimi.ai, model kimi-k3

  3. 03

    OpenRoutermoonshotai/kimi-k3, official pricing, full 1M context

  4. 04

    Wait for weights (July 27) — Hugging Face release; production needs 64+ accelerators

  5. 05

    Smoke-test — one API call to verify latency and billing before routing production traffic

Python · OpenAI-compatible API
from openai import OpenAI

client = OpenAI(
    api_key="YOUR_MOONSHOT_API_KEY",
    base_url="https://api.moonshot.ai/v1"
)

response = client.chat.completions.create(
    model="kimi-k3",
    messages=[
        {"role": "user", "content": "Analyze this codebase and identify performance bottlenecks..."}
    ]
)
07

Kimi K3 vs. The Competition

Use CaseBest PickWhy
Long, sustained coding sessionsKimi K3Leads SWE Marathon; 1M context
Complex bug fixes in large reposClaude Fable 5FrontierSWE advantage
Terminal/tool-heavy agent workflowsGPT-5.6 SolTerminal Bench leader
Multimodal document analysisKimi K3Best OmniDocBench; native vision + 1M
Cost-sensitive productionDeepSeek V4 ProOutput at $3.48/M
Open-source self-hosting (post 7/27)Kimi K3Most capable open weights
08

The Open-Source Promise: July 27

When weights drop July 27, Kimi K3 becomes the largest downloadable open-source model ever — the first above 2 trillion parameters. Trained with MXFP4 weights and MXFP8 activations for efficient inference. Day-0 support expected in transformers, vLLM, and SGLang.

Dates to bookmark: Right now → try at kimi.com · July 17–20 → WAIC Shanghai · July 27 → Hugging Face weights

09

Bottom Line

Kimi K3 is the most capable open-source AI model ever released. It doesn't win every benchmark — Claude Fable 5 and GPT-5.6 Sol still lead on specific tasks — but it's competitive across the board, outperforms them on long-horizon coding and document understanding, and ships with a 1M-token context at Sonnet-tier pricing. The July 27 open-weight release is the story to watch for the rest of 2026.

10

FAQ

Yes on kimi.com with a free account. API is pay-per-token at $3/$15 per 1M tokens.

Weights release July 27, 2026. Production needs 64+ H100-class accelerators — not a laptop LLM.

K3: 2.8T params, 1M context, stronger benchmarks. DeepSeek: far cheaper output ($3.48 vs $15 per 1M).

Yes for whole-codebase calls, long documents, and long-memory agents. Flat pricing makes full-window use practical.

Moonshot says subsequent updates. Currently only max is available at launch.

Closing note

K3's API is live today; self-hosting waits for July 27 and a GPU supernode most teams won't buy. If you're on Windows/Linux but need a real macOS GUI to run Kimi Code agents, handle keychain prompts, or benchmark against your current Claude/GPT stack, buying a Mac is expensive and SSH won't click system dialogs. VNCMac remote Mac rental gives you hourly VNC access to an isolated node — route K3 API traffic, run long-context coding sessions, and validate benchmarks from this article without a hardware commitment. See Mac rental plans to start.