# xAI Releases Grok 4.5, a Coding and Agent Model Trained With Cursor at Half Rivals' Price

The model prices at two dollars per million input tokens with a 500,000-token context window, and xAI claims parity with GPT-5.5 and near-Opus-4.8 coding scores it has not independently reproduced.

- Published: 2026-07-09T05:32:14.424Z
- Canonical: https://polylog.news/ai/2026-07-09/xai-releases-grok-4-5-a-coding-and-agent-model-trained-with
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [xAI](https://x.ai/news/grok-4-5), [MarkTechPost](https://www.marktechpost.com/2026/07/08/spacexai-releases-grok-4-5/)

xAI has released [Grok 4.5](https://x.ai/news/grok-4-5), which the company says is the first model in its lineup trained from the start for software engineering and autonomous agents rather than general conversation. It [shipped simultaneously inside Cursor](https://t.me/aipost/7461), the code editor whose real developer sessions xAI says it used as additional training data. That means the model was exposed to actual debugging, refactoring, and test-writing traces rather than synthetic code.

The commercial signal is the price. Grok 4.5 lists at [two dollars per million input tokens and six dollars per million output tokens](https://www.marktechpost.com/2026/07/08/spacexai-releases-grok-4-5/), with a 500,000-token context window and adjustable reasoning depth. That is far below what the leading closed models charge for output tokens. xAI's own figures claim performance on par with GPT-5.5 and approaching Anthropic's Opus 4.8 on coding, plus a first-place ranking on Harvey's Legal Agent Benchmark.

Those claims come from a vendor announcement, not from independent reproduction. GPT-5.5 and Opus 4.8 are the current closed-model leaders, and matching them at a third of the output price would be a real gain in capability per dollar rather than a single benchmark number. Independent tests on public agentic-coding suites, together with the questions about evaluation integrity raised elsewhere in today's edition, will determine whether the parity claim holds when third parties test it.

## What this means

The competitive pressure is on unit economics as much as on raw capability. By training on Cursor sessions and shipping inside the editor developers already use, xAI competes on distribution and cost at the same time, the two channels through which Anthropic and OpenAI earn revenue from coding. If the parity claim holds under independent testing, the labs charging three to five times more per output token on coding work would lose pricing power in exactly the workflow where usage volume is highest.

## What to watch

- Independent results on public agentic-coding benchmarks (SWE-Bench Verified, Terminal-Bench) run by parties other than xAI, which would confirm or undercut the GPT-5.5 parity claim.
- Whether Cursor keeps Grok 4.5 as a promoted default or treats it as one option among several, a signal of how much the editor values the co-training relationship versus model neutrality.
- Anthropic and OpenAI pricing moves on coding tiers in the next few weeks, which would indicate whether they regard the discount as real competitive pressure on demand.