Artificial Analysis launches Intelligence Index v4.2; GPT-6 Astra most token-efficient
ArtificialAnlys · x · 2026-09-05
Artificial Analysis announced Intelligence Index v4.2, accelerating parts of the planned v5 release to keep pace with the frontier. The new version has more complex, realistic tasks and more private test sets to prevent gaming, adding AA-Briefcase (a private agentic knowledge-work eval) and @HelloSurgeAI's GDP.pdf among others.
On token efficiency, GPT-6 Astra is more token-efficient than almost every other model near the intelligence frontier, while Claude Fable 5.1 and Gemini 3.8 Flash use the most tokens among models scoring at least 25 on the Index.
More from Models
- Altman: OpenAI sacrifices capability to preserve chain-of-thought monitoring — rohanpaul_ai · 2026-09-05
- GPT-6 default context window now 1.05M, but quality reportedly drops past 500K — bdsqlsz · 2026-09-05
- Codex repo leaks new "Persistent" reasoning effort: work until put to sleep — Singularitarian · 2026-09-05
- GPT 6 Astra voice mode rolls out, early users say it chats but won't work — oran_ge · 2026-09-05
- OpenAI Reports 'Persistent' Model Attack, Then Ships Persistent Mode to All Users — gleech · 2026-09-05
- ClaudeAI Weekly: ~17% usage cut coming Sept 13, watermarking live, new limit commands — ClaudeAI-mod-bot · 2026-09-05