Top AI Product

SpaceXAI pushed Grok 4.5 public on July 9, and Musk is calling it “Opus-class.” That’s a big claim for xAI’s new flagship model. The benchmarks are a split decision — but the token math isn’t.

What it is

A frontier reasoning-and-coding model, built on V9, a fresh 1.5-trillion-parameter base — roughly 3x the size of the Grok 4.3 foundation. It was trained alongside Cursor, the coding editor SpaceX bought for $60B, so real developer sessions (debugging traces, multi-file diffs, user corrections) went straight into training. Against Opus 4.8 on xAI’s own four benchmarks it went 2-2: 83.3% on Terminal-Bench 2.1, 62% on DeepSWE 1.0, losses on DeepSWE 1.1 and SWE-Bench Pro.

The API and why it matters

Live now in the SpaceXAI console, Grok Build, and across Cursor. Pricing is $2/$6 per million tokens, around 80 tokens/s. The real headline: on SWE-Bench Pro it averages 15,954 output tokens per task versus Opus 4.8’s 67,020. Same work, a quarter of the tokens. For anyone running coding agents at scale, that’s the number that pays the bill.

You Might Also Like

• Kimi k2 6 Beats gpt 5 4 and Claude Opus 4 6 on swe Bench pro
• Claude Sonnet 5 Scores 63 2 on swe Bench pro at a Third of Opus 4 8s Price
• Minimax m2 7 Scores 56 on swe pro and it Helped Build Itself
• Z ais glm 5 1 Tops swe Bench pro at 58 4 Trained on Zero Nvidia Hardware
• Mistral Medium 3 5 Scores 77 6 on swe Bench one Point shy of Gemini 3 1 pro