xAI
Grok 4.5
Version 4.5 benchmark runs across the shared Superbash visual prompts.
Official benchmark profile
How Grok 4.5 scores beyond our visual tests.
xAI describes Grok 4.5 as a coding, agentic-task, and knowledge-work model. xAI publishes different benchmark rows than OpenAI/Anthropic, so only direct source comparisons are shown.
DeepSWE 1.0
62.0%DeepSWE 1.1
53.0%SWE Marathon
29.0%Full official benchmark table6 rows with source settings and peer charts
| Benchmark | Area | Score | Setting / comparison |
|---|---|---|---|
| DeepSWE 1.0 | Coding | 62.0% | xAI reported software-engineering benchmark. |
| DeepSWE 1.1 | Coding | 53.0% | xAI reported software-engineering benchmark. |
| SWE Marathon | Long-horizon coding | 29.0% | xAI reported long-horizon coding benchmark. |
| SWE Bench Pro output tokens | Efficiency | 15,954 avg. | Average output tokens on SWE Bench Pro tasks. |
| Output-token efficiency | Efficiency | 4.2x fewer than Opus 4.8 | xAI comparison on SWE Bench Pro tasks.xAI compares Grok 4.5 against Opus 4.8 max on SWE Bench Pro tasks. |
| Serving speed | Latency | 80 TPS | Reported generation speed. |
Visual prompt runs
Benchmark runs
Open each generated scene, or compare the same prompt across models.
Hogwarts Broom Flight Simulator
Broom-flight scene testing depth, motion cues, castle scale, and fantasy mood.
Jabberwock
Dark fantasy encounter testing creature design, forest mood, and narrative staging.
Low Poly World
Stylized island build testing composition, color, and low-poly worldbuilding.
Office Life
Workplace vignette testing everyday scene logic, objects, and believable office detail.
Petri Dish
Microscopic ecosystem testing organic forms, scientific clarity, and cellular detail.
Universe Simulator
Cosmic system testing orbital structure, glowing bodies, scale, and simulation readability.
Vice City
Neon coastal city testing vehicles, architecture, atmosphere, and dense urban layout.