Top 5 AI Models Ranked by the Latest LLM Benchmarks – July 20, 2026

Updated: July 20, 2026

The race to build the world’s most capable artificial intelligence model continues to accelerate. New benchmark results show Anthropic, OpenAI, and Kimi competing at the top of the large language model rankings.

The following leaderboard compares five of the highest-ranked distinct AI model families on the Artificial Analysis Intelligence Index. The index combines multiple evaluations covering reasoning, mathematics, science, coding, knowledge, and agentic performance.

Latest Top 5 LLM Leaderboard

Claude Fable 5 Takes the Top Position

Anthropic’s Claude Fable 5 currently holds first place with an Intelligence Index score of 60. The tested configuration uses adaptive reasoning at maximum effort, with Claude Opus 4.8 available as a fallback model.

The model narrowly leads OpenAI’s GPT-5.6 Sol Max, which earned a score of 59. The one-point difference shows how closely matched the leading AI systems have become.

GPT-5.6 Sol Offers Strong Performance at a Lower Cost

GPT-5.6 Sol Max ranks second overall while recording a lower estimated evaluation cost than Claude Fable 5. It also supports a context window of approximately one million tokens, allowing it to process unusually large documents and conversations.

OpenAI also placed GPT-5.6 Terra Max among the five highest-ranked distinct model families. Terra Max recorded an Intelligence Index score of 55 and the fastest listed output speed in this comparison at 136 tokens per second.

Kimi K3 Ranks Third

Kimi K3 ranks third with an Intelligence Index score of 57. It also offers the largest context window in this group at approximately 1.05 million tokens.

The result places Kimi among the leading developers competing with Anthropic and OpenAI on advanced reasoning and knowledge benchmarks.

Claude Opus 4.8 Remains a Leading Model

Claude Opus 4.8 Max ranks fourth among the distinct model families included in this comparison, earning an Intelligence Index score of 56.

Although it has been overtaken by Claude Fable 5, Opus 4.8 remains one of the highest-performing models evaluated by Artificial Analysis.

What the Intelligence Index Measures

The Artificial Analysis Intelligence Index is a composite benchmark rather than a single test. Its current methodology combines evaluations designed to measure areas such as scientific reasoning, coding, professional tasks, factual knowledge, long-context reasoning, and autonomous computer use.

A higher Intelligence Index score indicates stronger combined performance across the included evaluations. However, benchmark scores do not measure every factor that may matter to users, including writing style, safety, reliability, application integrations, privacy, and performance on specialized tasks.

Important Benchmark Note

GPT-5.5 Xhigh also recorded an Intelligence Index score of 55, tying GPT-5.6 Terra Max. It was excluded from the table to keep the comparison limited to five distinct model entries.

AI benchmark rankings can change frequently as developers release new models and testing organizations update their evaluations. Readers should therefore treat this leaderboard as a snapshot dated July 20, 2026.

Sources

• Artificial Analysis LLM Leaderboard
• Artificial Analysis Intelligence Index