All tasks / Models

K2-Horizon-375B-A23B

MBZUAI IFM's open-weights K2 Horizon flagship MoE (Apache-2.0), 375B total, 23B active, 512K context.

MBZUAI Institute of Foundation Models
Open weights
Released 1 Sep 2026

Where it ranks

C
Chat & assistants
Models · Local · 31 on Artificial Analysis Intelligence Index (secondary benchmark)
C
Coding
Models · Local · 31 on Artificial Analysis Intelligence Index (secondary benchmark)
C
Workflow automation
Models · Local · 37.2% on AutomationBench-AA (Artificial Analysis)

Facts

Vendor: MBZUAI Institute of Foundation Models
Family: K2 Horizon
Released: 1 Sep 2026
Weights: Open: download and run it yourself
License: Apache-2.0
Parameters: 379B (23B active)
GPU memory to run locally: ~218 GB (4-bit ≈ 379 B × 0.57 GB + 1.5 GB ≈ 218 GB, estimate.)
Access: Free (open weights)
Last verified: 19 Oct 2026

Price from huggingface.co, 19 Oct 2026.

Benchmark results

Benchmark	Result	Entry on the leaderboard	As of	Source
Artificial Analysis Intelligence Index	31	`K2 Horizon 375B A23B` Raw value from the page's embedded model data; the site displays it rounded to an integer.	19 Oct 2026	artificialanalysis.ai
AutomationBench-AA (Artificial Analysis)	37.2%	`K2 Horizon 375B A23B` AutomationBench-AA Score (share of each task's objectives completed, a task with any guardrail violation scoring zero); raw value from the page's embedded model data (displayed rounded); rank 29 of 32; Tasks Completed 9.74%, Objectives Completed 66.04%; no last-updated date shown, so as_of = retrieval date.	19 Oct 2026	artificialanalysis.ai

Where you can use it

No tracked tool lists this model yet; as open weights it runs in local runtimes.

Sources