BREAKING
SpaceXAI Releases Grok 4.6
0
pts
AA Intelligence
0
K
Context tokens
Output Costs: $6 vs $30 vs $50
Grok 4.6
6
GPT-5.6 Sol
30
Fable 5
50
A Jagged Benchmark Lead
Grok 4.6
Leads
●
GDPVal-AA
●
Harvey LAB
Rival strengths
Mixed
●
Fable leads CursorBench
●
Sol leads DeepSWE and Terminal-Bench
Long-Horizon Training Pipeline
1
Supplemental training
↓
2
Refresh SFT traces
↓
3
Filter weak traces
↓
4
Agentic RL
Real Work Is the Next Test
AI NEWS BLITZ
SpaceXAI has released Grok 4.6 for long-running agent and coding work.