Four Frontier Launches In Eight Days Six Labs Now Field A Model Above 50 On The Artificial Analysis Intelligence Index

Several labs now offer near-frontier models at much lower per-task cost, such as GPT-5.6 Sol at $1.04 against $2.75 for Fable 5. Revisit your default model and routing choices.

• Artificial Analysis reported that six labs exceeded 50 on its Intelligence Index by July 17, 2026, up from two in early June. The top three models came from different labs and were separated by only three points.
• In the July 2026 comparison, GPT-5.6 Sol scored one index point below Claude Fable 5 at $1.04 versus $2.75 per task. Near-frontier capability had become available at a substantially lower measured task cost.
• The July 2026 Kimi K3 results differed by task. Artificial Analysis placed it third on GDPval-AA v2 but second on its private AA-Briefcase knowledge-work benchmark, illustrating why an aggregate rank is not a task-specific verdict.

• benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.

• articleKimi K3 1 Pro Launch Benchmarks
• articleHealthbench Professional Health Conversations
• articleAa Personalitybench
• articleAnnouncing Aa Lcr

• postLast week the Intelligence Index vs Cost Pareto frontier moved out substantially.ArtificialAnlys
• postMuse Spark 1.2 is among the most cost-efficient models at its intelligence…Artificial Analysis
• postNew releases reshape the Intelligence Index cost/performance frontierArtificial Analysis

Checking sign-in…

Loading comments…