SpaceXAI: Grok 4.5
by xAItext+image+file->text4 endpoints
Neura Intelligence Index
85.3/ 100
Rank
#22—
Confidence
lowBenchmarks
2(5% coverage)
Overview
Grok 4.5 is a model from SpaceXAI with frontier performance on coding, knowledge work, and STEM.
Capabilities
Text generation
Image understanding
Tool use / Function calling
Structured output
Extended reasoning
Prompt caching
Long context
File input
Top-P sampling
Deterministic seed
Log probabilities
Modalities
Input
TextImageFile
Output
Text
Technical Specifications
Context Window
500K tokens
Max Output
450K tokens
Tokenizer
Grok
Uptime (24h)
9969.2%
Country
🇺🇸US
Release Date
July 16, 20262mo
Supported Parameters
include_reasoninglogprobsmax_tokensreasoningreasoning_effortresponse_formatseedstructured_outputstemperaturetool_choicetoolstop_logprobstop_p
Benchmark Performance
coding
| Benchmark | Score | Source |
|---|---|---|
| SWE-Bench Pro | 64.7 | verified |
reasoning
| Benchmark | Score | Source |
|---|---|---|
| GPQA Diamond | 93 | verified |
Performance Over Time
Pricing
Per 1M tokens
Input$2.00
Output$6.00
Cache read$0.30
Added
July 8, 2026
Last synced 9/18/2026