K2 Horizon

K2 Horizon

Institute of Foundation Models
+
+

Related Products

  • LTX
    182 Ratings
    Visit Website
  • LM-Kit.NET
    29 Ratings
    Visit Website
  • Gemini Enterprise Agent Platform
    999 Ratings
    Visit Website
  • InEight
    136 Ratings
    Visit Website
  • Zendesk
    7,958 Ratings
    Visit Website
  • FinOpsly
    3 Ratings
    Visit Website
  • Creatio
    586 Ratings
    Visit Website
  • Daylight
    11 Ratings
    Visit Website
  • NetBrain
    285 Ratings
    Visit Website
  • Checksum.ai
    1 Rating
    Visit Website

About

GLM-5.3-Flash is Z.ai’s natively multimodal model in the GLM-5 series (previously previewed as Ox Alpha), designed to deliver strong coding, agentic, visual, and knowledge-work performance at relatively low inference cost. It uses 320 billion total parameters with 18 billion active parameters, along with a hybrid architecture that combines sparse and linear attention to reduce the cost of long-context processing. The model supports context lengths of up to one million tokens and was trained on a 30-trillion-token multimodal corpus. GLM-5.3-Flash can reason across text, images, documents, interfaces, dashboards, and other visual information while using that feedback to refine its own outputs. Z.ai reports substantial gains over GLM-5.2 on coding and agentic benchmarks, including DeepSWE and AutomationBench, while approaching higher-cost frontier models on several evaluations.

About

K2 Horizon is a connected fleet of six open models spanning 375B-A23B, 36B-A4B, 32B, 7B, 3.7B, and 0.9B, designed to deliver strong performance across reasoning, mathematics, coding, agentic tasks, and general capabilities. The models share core architecture, vocabulary, training methodology, interfaces, evaluation infrastructure, and deployment tooling, making it easier to move between sizes and route workloads dynamically. The 375B-A23B model is the fleet’s most capable option for complex reasoning, software engineering, research, and long-horizon agentic work, while the 32B and 36B-A4B models target powerful local deployment. The 36B-A4B model introduces Mixture-of-Value Attention, combining sparse attention with Mixture-of-Experts layers to activate about 4 billion parameters per token while approaching the performance of the dense 32B model.

Platforms Supported

Windows
Mac
Linux
Cloud
On-Premises
iPhone
iPad
Android
Chromebook

Platforms Supported

Windows
Mac
Linux
Cloud
On-Premises
iPhone
iPad
Android
Chromebook

Audience

Developers, AI engineers, agent builders, researchers, and organizations that need cost-efficient multimodal reasoning, long-context processing, advanced coding, visual analysis, and autonomous workflow capabilities

Audience

AI researchers, developers, and organizations searching for open models for reasoning, coding, tool use, agentic workflows, and deployments

Support

Phone Support
24/7 Live Support
Online

Support

Phone Support
24/7 Live Support
Online

API

Offers API

API

Offers API

Screenshots and Videos

Screenshots and Videos

Pricing

$0.15 per 1M tokens (input)
Input: $0.15 per 1M tokens
Output: $0.50 per 1M tokens
Cached input: $0.03 per 1M tokens
Free Version
Free Trial

Pricing

No information available.
Free Version
Free Trial

Reviews/Ratings

Overall 5.0 / 5
ease 5.0 / 5

Reviews/Ratings

Overall 0.0 / 5
ease 0.0 / 5
features 0.0 / 5
design 0.0 / 5
support 0.0 / 5

This software hasn't been reviewed yet. Be the first to provide a review:

Review this Software

Pros & Cons from Real Users

Pros

  • What makes it exciting is that it seems built for the exact workloads developers care about right now: long-horizon coding, complex reasoning, big-context analysis, and agentic workflows. A million-token context window is especially useful if you want to drop in a large repo, long spec, research corpus, or messy project history and have the model reason across it.

Cons

  • I would treat it as something exciting to test, not something to blindly trust with sensitive work. Even the independent Ox Alpha site warns that messages are processed by the upstream model API, so I would keep secrets, private code, and customer data out of it until there is a clearer owner, model card, privacy policy, and production story.

Training

Documentation
Webinars
Live Online
In Person

Training

Documentation
Webinars
Live Online
In Person

Company Information

Z.ai
Founded: 2019
China
z.ai

Company Information

Institute of Foundation Models
Founded: 2025
United States
ifm.ai/blog/k2/

Alternatives

Alternatives

LongCat-2.0

LongCat-2.0

LongCat
MiMo-V2.5-Pro

MiMo-V2.5-Pro

Xiaomi Technology
GPT-5.6 Sol

GPT-5.6 Sol

OpenAI
Ling 3.0 Flash

Ling 3.0 Flash

Ant Group
MiniMax M3

MiniMax M3

MiniMax
Command A+

Command A+

Cohere AI
Qwen3.5

Qwen3.5

Alibaba
Hy4

Hy4

Tencent

Categories

Categories

Integrations

Cheaper Inference
Claude Code
DeepSeek Harness
GLM Coding Plan
Hermes Agent
OpenClaw
OpenCode Go
OpenCode Zen
OpenRouter
Pi Agent
Z.ai
omp

Integrations

Cheaper Inference
Claude Code
DeepSeek Harness
GLM Coding Plan
Hermes Agent
OpenClaw
OpenCode Go
OpenCode Zen
OpenRouter
Pi Agent
Z.ai
omp
Claim GLM-5.3-Flash and update features and information
Claim GLM-5.3-Flash and update features and information
Claim K2 Horizon and update features and information
Claim K2 Horizon and update features and information