News & Updates, Code Editors

Cursor and SpaceXAI Release Grok 4.5, Their First Joint Model

SpaceXAI and Cursor have jointly launched Grok 4.5, a 1.5-trillion-parameter coding model available now across all Cursor plans at $2/$6 per million tokens — weeks before their $60B acquisition deal closes.

3 min read
Cursor and SpaceXAI Release Grok 4.5, Their First Joint Model

Image by SpaceX

Cursor and SpaceXAI Release Grok 4.5, Their First Model Trained Together

Cursor and SpaceXAI jointly launched Grok 4.5 on July 8, the first AI model the two companies have built together. The release is available now across all Cursor plans and from the SpaceXAI console, weeks before SpaceX's $60 billion acquisition of Cursor formally closes.

SpaceX announced the acquisition of Cursor's parent company Anysphere in mid-June at a $60 billion valuation. The deal is expected to close in Q3 2026. In the meantime, SpaceXAI — the AI unit that emerged when SpaceX absorbed xAI in February — and Cursor have moved ahead with a joint model trained on Cursor's development workflow data and SpaceXAI's Colossus compute infrastructure.

What Grok 4.5 Is

Grok 4.5 is a mixture-of-experts model built on a 1.5-trillion-parameter V9 foundation. SpaceXAI describes it as an Opus-class model designed primarily for software engineering, legal work, financial analysis, and agentic tasks — rather than general-purpose chat. Elon Musk framed it on X as "an Opus-class model, but faster, more token-efficient and lower cost."

On coding benchmarks, SpaceXAI's evaluations put Grok 4.5 at 83.3% on Terminal-Bench 2.1, competitive with Fable 5 at 84.3% and GPT-5.5 at 83.4%. On SWE Marathon pass@1, it scores 29.0%, ahead of Opus 4.8 at 26% and Fable 5 at 24%. SWE Bench Pro resolution rate comes in at 64.7%, behind Fable 5 at 80.4% and Opus 4.8 at 69.2%. These are vendor-run benchmarks and independent verification is pending.

The pricing is the sharpest differentiator. Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens. A faster premium variant runs at $4/$18. By comparison, Claude Opus 4.8 is priced at $5 input/$25 output, and GPT-5.6 Sol at $5/$30. SpaceXAI is offering free Grok 4.5 usage for a limited period in Grok Build and Cursor.

What Changes for Cursor Users

Grok 4.5 is available immediately in Cursor across desktop, web, iOS, CLI, and the Cursor SDK on all plans. The model is live in Grok Build and the SpaceXAI console via API key. EU availability is expected in mid-July.

For developers already working inside Cursor, Grok 4.5 sits alongside Composer 2.5 — Cursor describes the two as different model weight classes aimed at different task profiles. The key differentiator is that Grok 4.5 was trained with data from Cursor's real developer workflows, agentic sessions, and multi-file editing traces — meaning this is the first model specifically shaped by how Cursor users actually code rather than a general-purpose model integrated as a third-party option.

SpaceXAI has also committed to shipping new foundation models roughly monthly through the end of 2026. Grok 4.5 is the first public proof point of that cadence and the first to carry Cursor data as a co-development input rather than a downstream integration.

What the Acquisition Means for the Product

The Cursor acquisition has not yet formally closed, but the joint model launch makes the product relationship concrete before the corporate one is finalized. Reports indicate Grok 4.5 was trained partly using the same compute capacity SpaceXAI simultaneously leases to Anthropic and Google — a detail that underscores the infrastructure leverage SpaceXAI is building into its competitive position.

For Cursor users, the practical question is how the SpaceX acquisition will affect model availability, pricing, and product independence over time. The current message from both companies is continuity: all plans get Grok 4.5 access, existing settings carry over, and the model is pitched on price-to-performance rather than exclusivity. Whether that holds after the deal closes in Q3 is what developers should track closely.

Share:

Other Latest News

OpenAI Releases GPT-5.6 and GPT-Live to All Users Today
News & Updates, AI Agents, Voice

OpenAI Releases GPT-5.6 and GPT-Live to All Users Today

OpenAI publicly launches its most powerful model family — Sol, Terra, and Luna — today alongside GPT-Live, a new full-duplex voice architecture that replaces Advanced Voice Mode across all ChatGPT tiers.

Jul 9, 2026
Anthropic Adds Model Controls and Spend Alerts to Claude Enterprise
AI Agents, News & Updates

Anthropic Adds Model Controls and Spend Alerts to Claude Enterprise

Anthropic's new Claude Enterprise admin controls give platform admins per-user cost breakdowns, model-level entitlements, and real-time spend-threshold alerts — targeting the budget blowouts that have quietly burned enterprise AI teams in 2026.

Jul 6, 2026
Gemini Spark Comes to Mac With Local File Access and MCP Support
AI Agents, News & Updates

Gemini Spark Comes to Mac With Local File Access and MCP Support

Google has added Gemini Spark to the Gemini desktop app for macOS, giving it access to local files, MCP server connections, real-time topic tracking, and new integrations with Canva, Dropbox, and more.

Jul 3, 2026
Google Drops ADK 2.0 and Genkit Agents for AI App Builders
AI Agents, News & Updates

Google Drops ADK 2.0 and Genkit Agents for AI App Builders

Google published three developer-facing launches in 24 hours: ADK 2.0's graph-based workflow runtime, the Genkit Agents API for full-stack conversational AI, and a Google Cloud Workbench VS Code extension — a coordinated push to own the agent-development stack.

Jul 3, 2026
Fable 5 Returns Globally as US Lifts Export Controls on Anthropic
News & Updates, AI Agents, Security

Fable 5 Returns Globally as US Lifts Export Controls on Anthropic

Anthropic restored global access to Claude Fable 5 on July 1 after the US Commerce Department lifted the 19-day export controls. Claude Code and Claude.ai users get the model back with a new 99%+ jailbreak classifier and an industry-wide safety framework now in development.

Jul 2, 2026
Anthropic Launches Claude Sonnet 5 With Near-Opus Coding Performance
AI Agents, News & Updates

Anthropic Launches Claude Sonnet 5 With Near-Opus Coding Performance

Anthropic's new Claude Sonnet 5 brings near-Opus 4.8 agentic coding performance at Sonnet pricing, with a 1M-token context window and introductory rates of $2/$10 per million tokens through August 31.

Jul 1, 2026
← Scroll for more →