Grok 4.7 Costs $8. Claude and GPT-6 Cost $60. Guess Who Scores Higher.
xAI (now branded SpaceXAI) released Grok 4.7 on Sept 21, 2026, priced at $8 for 1M input + 1M output tokens combined — roughly a seventh of the $60 Claude's Fable 5.1 and OpenAI's GPT-6 Astra charge for the same mix. The catch: Grok 4.7 scores well behind both, landing at 46 on the Artificial Analysis Intelligence Index versus 53 for Claude and GPT-6 (tied), and 26% on Terminal-Bench 4.0 versus GPT-6 Astra's 60%.
Receipts
- primaryhttps://x.ai/news/grok-4-7
- alsohttps://the-decoder.com/xai-launches-grok-4-7-at-bargain-prices-but-benchmarks-reveal-a-wide-gap-to-claude-and-gpt-6/
- alsohttps://kingy.ai/blog/grok-4-7-benchmarks-specs-frontier-comparison/
- alsohttps://www.theneuron.ai/news/xais-grok-47-makes-frontier-ai-a-price-war/
- alsohttps://xenospectrum.com/en/xai-grok-4-7-pricing-copilot/
Flagged:Primary is xAI's own announcement (x.ai/news/grok-4-7, now published under the SpaceXAI brand), which covers the release, specs and same-as-4.6 pricing but not the cross-lab comparison. The comparative benchmark figures ($8 vs $60 combined 1M-token cost, 26% vs 60% on Terminal-Bench 4.0, 46 vs 53 on the Artificial Analysis Intelligence Index) are the-decoder's own analysis, independently corroborated by kingy.ai and The Neuron. The AA Intelligence Index is the same index this site's own Leaderboard (src/data/leaderboard.ts) tracks for every lab, which is why the card and captions point readers there. Claude Fable 5.1 and GPT-6 Astra tie at 53 on that index; Grok 4.7 sits at 46, mid-pack. This is a benchmark comparison, not a recommendation — which model is 'better' depends on the task and budget.