Skip to content

Language models

Intelligence, value, output speed, and task-specific benchmark leaders.

Snapshot: September 20, 2026

Analysis

Math

Artificial Analysis Math Index · Higher is better

Eligible
0
Candidates
653
Coverage
0.0%

unavailable

No eligible math_index observations are available for this run.

Output speed

Median output tokens per second · Higher is better

Eligible
329
Candidates
653
Coverage
50.4%
Rank Model Creator Median output tokens per second Since prior snapshot
1 Celeris-1 Celeris 1,630.16 Not comparable
2 Mercury 2 Inception 413.81 Not comparable
3 Gemini 2.5 Flash-Lite (Reasoning) Google 412.96 Not comparable

Overall intelligence

Artificial Analysis Intelligence Index · Higher is better

Eligible
644
Candidates
653
Coverage
98.6%
Rank Model Creator Artificial Analysis Intelligence Index Since prior snapshot
1 Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) Anthropic 53.4 Not comparable
2 Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback) Anthropic 53.2 Not comparable
3 GPT-6 Astra (max) OpenAI 52.7 Not comparable

Token-price value

Intelligence Index per local 3:1 blended token price · Higher is better

Eligible
409
Candidates
653
Coverage
62.6%
Rank Model Creator Intelligence Index per local 3:1 blended token price Since prior snapshot
1 Agnes 3.0 Flash Sapiens AI 473.333 Not comparable
2 Llama 3.1 Instruct 8B Meta 250.909 Not comparable
3 Agnes 2.5 Pro Beta Sapiens AI 234.667 Not comparable

Data source: Artificial Analysis .