SpaceXAI releases Grok 4.5 at $2/$6 per million tokens, trained with Cursor on NVIDIA GB300 GPUs
SpaceXAI has released Grok 4.5, the first model developed after its rebrand from xAI and trained in partnership with Cursor. The model is priced at $2 per million input tokens and $6 per million output tokens, and was trained across tens of thousands of NVIDIA GB300 GPUs focused on coding, science, engineering, and math datasets.
Grok 4.5 — Quick Specs
SpaceXAI releases Grok 4.5 at $2/$6 per million tokens, trained with Cursor on NVIDIA GB300 GPUs
SpaceXAI has released Grok 4.5, its first model after rebranding from xAI and the first trained in partnership with AI coding company Cursor. The model is priced at $2 per million input tokens and $6 per million output tokens.
Training infrastructure and focus
According to SpaceXAI, Grok 4.5 was trained across tens of thousands of NVIDIA GB300 GPUs on datasets focused on coding, science, engineering, and math. The company designed the model specifically "to excel at coding, agentic tasks and knowledge work."
SpaceXAI claims Grok 4.5 outperforms other leading models at real engineering tasks and can create functional applications with minimal instructions. The company demonstrated the model generating an interactive solar system simulation from a single prompt.
Pricing comparison
Grok 4.5's pricing positions it between OpenAI's GPT-5.6 variants:
- Grok 4.5: $2 input / $6 output per million tokens
- GPT-5.6 Sol (most powerful): $5 input / $30 output per million tokens
- GPT-5.6 Luna (most affordable): $1 input / $6 output per million tokens
SpaceXAI claims Grok 4.5 responds faster than "flash" models while delivering results "at far lower costs," though specific benchmark comparisons were not provided.
Availability and integration
Grok 4.5 is now the default model for Grok Build, SpaceXAI's terminal-based AI coding agent that handles coding, Excel, PowerPoint, and Word tasks. The model is available through the SpaceXAI console and integrated into all Cursor plans.
The model is not yet accessible in the European Union, with availability expected in mid-July.
SpaceXAI-Cursor partnership details
In April, SpaceXAI and Cursor announced a partnership to develop AI together. The deal structure involves either a $10 billion investment by SpaceXAI into Cursor, or a full acquisition of Cursor for $60 billion "later this year." The companies have not yet disclosed which option they will pursue.
Cursor stated on X: "We've partnered with SpaceXAI to train Grok 4.5. It's our most powerful model yet and the first we've built for more than software engineering."
What this means
Grok 4.5 represents the first concrete output from the SpaceXAI-Cursor partnership and signals SpaceXAI's focus on coding and engineering applications post-rebrand. The pricing undercuts OpenAI's premium GPT-5.6 variant significantly, though it matches the output pricing of OpenAI's budget option. The model's performance claims and "flash" speed assertions require independent verification through benchmarks, as SpaceXAI has not yet published standardized evaluation scores. The pending $10-60 billion deal between SpaceXAI and Cursor makes this release a critical test of their collaborative capabilities.
Related Articles
OpenAI Halts Parts of Astra Model Development After It Hit 'Critical' Cybersecurity Threshold
OpenAI disclosed that its in-development Astra model showed cyberattack capabilities strong enough that it cannot rule out a 'Critical' risk classification. The company has paused related internal activity and added security controls under its Preparedness Framework.
Mistral's 3B-Parameter Shieldstral Matches 20B Safety Model on Text Benchmarks
Mistral's new Shieldstral, a 3-billion-parameter open-weight safety classifier, posts an 84.9% F1 score on text benchmarks—tying OpenAI's GPT-OSS-Safeguard-20B, a model roughly seven times larger. The model lets operators define safety rules at runtime using plain-language yes/no questions instead of fixed taxonomies.
Mistral AI Releases Shieldstral-1.0-3B, a 3B-Parameter Policy-Adaptive Safety Classifier
Mistral AI has released Shieldstral-1.0-3B, a compact open-weight safety classifier that evaluates text and images against natural-language policies specified at inference time. The 3B model runs on a single GPU and reports F1 scores competitive with or exceeding larger moderation models like LlamaGuard-4-12B and GPT-OSS-Safeguard-20B on multiple benchmarks.
Black Forest Labs Launches FLUX 3 Video, Claims It Beats Seedance 2.0 on Elo Rankings
Black Forest Labs has made FLUX 3 Video generally available via its API, offering up to 20-second HD/Full HD clips with native audio and lip-sync in 14+ languages. The company claims its internal Elo benchmarks put the model ahead of Seedance 2.0, Gemini Omni Flash, and Minimax H3.
Comments
Loading...