Grok 4.5

Grok 4.5 is SpaceXAI's flagship model designed to excel at coding, agentic workflows, and professional knowledge work. It was trained on datasets spanning coding, science, engineering, and mathematics, with a particular emphasis on datasets developed alongside Cursor to strengthen real-world software engineering capabilities. The model combines intelligent reasoning with efficient reasoning modes, making it adaptable to tasks that range from lightweight assistance to deeper, multi-step problem solving.

Independent benchmark testing on Snorkel's GDPval+ dataset, which contains expert-authored professional workplace tasks across the U.S. economy, showed Grok 4.5 delivering the strongest overall performance compared to other frontier models evaluated in the same study. On engineering-focused evaluations, including the DeepSWE agentic coding suite, Grok 4.5 posted competitive pass@1 scores against leading competitors, and it achieved the highest resolution rate on SWE Marathon. Practically, Grok 4.5 is a strong fit for teams building agentic coding assistants, automated engineering pipelines, and tools that need to reason over structured professional deliverables such as documents, spreadsheets, and presentations.

Vercel AI Gatewayspacexai/grok-4.5grok

Open provider page Provider docs

Quick Info

Provider: Vercel AI Gateway
Model key: spacexai/grok-4.5
Release date: Jul 8, 2026
Last updated: Jul 8, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $2.00
Output token cost: $6.00

Limits

Output tokens: 500,000 tokens
Context window: 500,000 tokens

Transparent token rates

Compare Grok 4.5 pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about Grok 4.5

Vercel AI Gateway

Coverage

xAI's official news index lists a direct Grok 4.5 entry dated July 28, 2026: "Grok 4.5, our smartest coding model, is now available in GitHub Copilot," confirming the model as a distinct xAI release positioned for coding workflows. The same page's August 14, 2026 entry introducing Grok 4.6 explicitly states it "builds Beyond the Copilot availability line, the xAI news excerpt situates Grok 4.5 within a broader product lineup that includes Grok Bot, Build Mode, Imagine Image 2.0, Imagine Video 1.5, and Grok Voice Think Fast 2.0, providing useful family-level context for where Grok 4.5 sits relative to xAI's other releases in mid-to-l

Vercel AI Gateway

Coverage

FullStack Labs' third-party explainer covers xAI's Grok 4.5 as a model aimed at "software engineering, agents, and long-form analysis," with the article framing it for teams evaluating Claude, GPT, and cost tradeoffs. The supplied excerpt states SpaceXAI announced Grok 4.5 on July 8, 2026 as its "most capable model so The explainer describes Grok 4.5's technical orientation as long-context reasoning, tool use, and stable behavior over real repositories and internal knowledge sources, with practical capabilities including reading and modifying large codebases, coordinating multi-step plans through tools or APIs, and processing long r

Vercel AI Gateway

Coverage

Tech Insider Ireland's launch coverage states that xAI released Grok 4.5 on July 8, 2026 with pricing of $2 per million input tokens and $6 per million output tokens, a level the article frames as undercutting Anthropic's Claude Opus 4.8 by a significant margin. Three initial distribution channels are named: Grok Build Beyond the launch-day facts, the article tracks a staged regional rollout: AIWiz-reported EU API access on July 17, 2026, followed by consumer surfaces (Grok.com, X, iOS, Android) on July 22, 2026, with xAI positioning developers and IDE workflows ahead of the consumer chatbot. The model is described in xAI's documenta