K2 Think V2

Name: K2 Think V2
Author: MBZUAI

Released

2026-01-27

Last refreshed

2026-09-10

Status

Researched 149d ago

Open sourceCommercial use: permittedLong context

K2 Think V2 is a released long context model with open-source and 512K context; evaluate it while provider pricing coverage matures.

MBZUAI releases · 7 in the last 12 months · this family litChangelog →

Specifications

Family: K2 Think
Released: 2026-01-27
Context: 512K
Parameters: 70B
Architecture: Decoder Only
Openness: Open source
License: Apache 2.0OSI-approvedCommercial use: permitted
Weights: Unknown
Code: Unknown

Created by

MBZUAI

Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) is the world's first graduate-level, research-focused A

Abu Dhabi, UAE

Founded 2019

Website

Pricing

No tracked provider pricing is available yet.

About

K2 Think V2 is a 70B parameter open-source reasoning model from MBZUAI (Mohamed bin Zayed University of Artificial Intelligence), developed in collaboration with G42 and Cerebras Systems. Built on the K2-V2 base model, it is optimized for deep multi-step reasoning across math, science, coding, logic, and long-context tasks with a 512K token context window. Released under a '360-open' approach including weights, training code, data, and evaluation tools.

Providers

No active provider routes are tracked for this model yet.

Capabilities

No model capability flags are currently sourced.

Benchmark peer barsfor Long context

No task-mapped benchmark peers are available for this model yet.

Benchmark scores(1)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.

Benchmark	Score	Version	Evaluation	Source
Google-Proof Q&A	71.3	diamondObserved 2026-04-18	—	Source

Migration checks

No linked migration route is available for this model yet.