Grok 4.5

Coding modelGrok 4

Grok 4.5

Release date: 2026-07-08Updated: 2026-07-13 15:23:34.785Knowledge cutoff: 2026-02-011,167

Live demoGitHubHugging FaceCompare

Parameters

Not disclosed

Context length

500K

Chinese support

Not supported

Reasoning ability

Grok 4.5 是 SpaceXAI 于 2026 年 7 月 8 日发布的旗舰模型，面向编程、Agentic 任务和知识工作，支持文本与图像输入、500K 上下文、函数调用、联网搜索、代码执行与 low/medium/high 三档推理；知识截止日期为 2026 年 2 月 1 日。

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Grok 4.5

Model basics

Reasoning traces

Supported

Thinking modes

Thinking Level · High (Default)Thinking Level · LowThinking Level · Medium

Context length

500K tokens

Max output length

No data

Model type

Coding model

Modality (in / out)

Text, Image → Text

Release date

2026-07-08

Model file size

No data

MoE architecture

Total params / Active params

No data / N/A

Knowledge cutoff

2026-02-01

Grok 4.5

Open source & experience

Code license

不开源

Weights license

不开源

GitHub repo

GitHub link unavailable

Hugging Face

Hugging Face link unavailable

Live demo

https://grok.com/

Grok 4.5

Official resources

Paper

Introducing Grok 4.5

DataLearnerAI blog

No blog post yet

Grok 4.5

API details

API speed

4/5

💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.

Learn about pricing modes

Standard

Type	Condition	Input	Output
Text	-	$2.00/ 1M	$6.00/ 1M

Cache PricingPrompt Cache

Type	TTL	Write	Read
Text	-	-	$0.500/ 1M

Grok 4.5

Benchmark Results

Grok 4.5 currently shows benchmark results led by SWE-Bench Pro - Public (4 / 48, score 64.70), TerminalBench 2.1 (7 / 21, score 83.30), DeepSWE (9 / 16, score 53). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.