Claude Fable 5.1Anthropic · max · fallback
53Intelligence Index
AI Model Rankings
Live model trends and rankings based on usage across XAPI, showing what developers are building with now. View all models
Usage data through 2026-09-16Preview data shown on this page
INDEPENDENT BENCHMARK
Frontier model ranking
Six leading models from the Artificial Analysis Intelligence Index v4.3. The index combines difficult evaluations; it is not a percentage of intelligence. The leading score in this edition is 53. These scores are separate from the XAPI usage preview below.
GPT-6 AstraOpenAI · max
53Intelligence Index
Claude Opus 5Anthropic · max
51Intelligence Index
Claude Fable 5Anthropic · fallback
50Intelligence Index
Muse Spark 1.3Meta · max
48Intelligence Index
GPT-5.6 SolOpenAI · max
47Intelligence Index
Source: Artificial Analysis Intelligence Index v4.3 · September 7, 2026
Top models
Weekly model usage across XAPI
GPT-5.4DeepSeekClaudeOthers
Leaderboard
Models ranked by tokens processed
| # | Model | Processed | Change |
|---|---|---|---|
| 1. | gpt-5.4OpenAI | 8.42B | 18% |
| 2. | deepseek-flashDeepSeek | 7.16B | 34% |
| 3. | claude-sonnet-4.6Anthropic | 5.84B | 12% |
| 4. | gemini-3.1-proGoogle | 4.91B | 4% |
| 5. | glm-5.3-flashZhipu AI | 3.76B | 27% |
Top models by task
Leading models for common workloads
1claude-sonnet-4.6Anthropic42%
2gpt-5.4OpenAI31%
3deepseek-flashDeepSeek19%
1gpt-5.4OpenAI42%
2gemini-3.1-proGoogle31%
3claude-sonnet-4.6Anthropic19%
1claude-sonnet-4.6Anthropic42%
2gemini-3.1-proGoogle31%
3gpt-5.4OpenAI19%
1deepseek-flashDeepSeek42%
2gpt-5.4OpenAI31%
3qwen3.5-maxAlibaba19%
Capability and price benchmarks
Capability score versus input price per million tokens
IntelligenceInput price
Fastest models
Available models ranked by output speed
1glm-5.3-flash1M Context
142tokens/sec
2deepseek-flash1M Context
126tokens/sec
3kimi-k2.5256K Context
118tokens/sec
4qwen3.5-max1M Context
109tokens/sec
5gemini-3.1-pro2M Context
83tokens/sec
6gpt-5.41M Context
78tokens/sec
How rankings are measured
What is measured
Input and output tokens routed through XAPI are aggregated by model version.
Selection period
Today, week, and month cover the latest complete 1, 7, and 30-day windows. Trending compares adjacent weeks.
What rankings mean
Usage reflects adoption on XAPI. It does not establish model quality, total market share, or an official benchmark result.