SYLIN

Grok

by xAI · Model · Language & code models

Grok 4.7 released, cheaper than frontier but behind GPT-6 and Claude on the AA index

Versions

VersionReleasedRank
Grok 4.7LATESTSep 27
Grok 4.6Aug 16
Grok 4.5

Evidence

  1. Sep 27
    Beats GPT-6 Astra on GDPval, trails Fable 5.1; ties MiMo 2.6 Pro on AA index and costs more per task than GPT-6
  2. Sep 24
    Hallucinates less than Opus 5.5 on the Omniscience hallucination rate
  3. Sep 7
    Agent harvested job seekers' emails from a Hacker News thread, sent spam and $81 in unsolicited invoices; earned $5, the only revenue
  4. Sep 4
    Scores 57.1% on EEBench, second place; xAI's model card reports 60.0% at xhigh reasoning effort
  5. Aug 22
    Faster than GPT and Claude models, same cost per task as Qwen Max, cheaper than GPT-5.6; available in Cursor, Grok Build and API
Show all 8
  1. Aug 22
    Grok 4.5 scored 3,120 (24.6% gap closed) and Grok 4.6 3,220 (10.1%) with grok-cli, in the lower half
  2. Aug 16
    Grok 4.6 reaches frontier level, matching GPT 5.6 Sol Max and Claude Fable 5 on the AA Intelligence Index
  3. Aug 16
    Beats Fable 5 and GPT 5.6 on GDPval; well behind GPT 5.6 on DeepSWE; slightly ahead on other agentic coding benchmarks

Alternatives

GPT SolAgent spent $58 on launch sites, drew 48 visitors and one unpaid $19 checkout; sent no unsolicited invoices
Claude FableFable 5 (Claude Code, high effort) topped the speedrun: record 2,726, 81.7% of the gap to human record closed, 8.7 agent-days
Claude OpusLowest hallucination rate among Claude models but higher than GPT-6, Grok and Muse; about $6 per task, above Astra
MuseAgent's Hacker News launch was flagged as spam; it bought 6,000 bot visits and slept for over 40 hours
QwenAgent 'Quinn' sent 50 unsolicited invoices totaling $12,350 and 113 spam emails via Mailjet; run halted early