BREAKING
xAI Debuts Grok 4.5 for Coding
0
k
context tokens
0
TPS
inference speed
API Pricing per 1M Tokens
Input
2
Output
6
How Grok 4.5 Was Built
1
GB300 GPU training
↓
2
RL on SWE tasks
↓
3
Cursor co-training
Strengths and Caveats
Praise
Positive
●
Beat Claude and GLM on tasks
●
Durable in agentic loops
●
Office plugin support
Caveats
Watch
●
Quotas consumed quickly
●
Value vs top tiers unproven
Grok 4.5 Targets Cost and Speed
AI NEWS BLITZ
xAI has released Grok 4.5, a new flagship model tuned for coding and agentic tasks.