Preferences and analytics

BenchSift keeps your language, theme, and comparison preferences in your browser. Audience measurements also help improve the site.

Back

Alibaba

Qwen3.8 Max

40.3 Intelligence Index2026-08-03ReasoningClosed weights
Artificial Analysis Indices
Default
Intelligence40.3
Coding71.8
Agentic49.6
Standard Benchmarks
GPQA Diamond92.7%
HLE43.0%
SciCode53.2%
AA-LCR78.3%
Terminal-Bench 2.181.3%
τ-Bench Banking51.3%
Capabilities
Context window1.0M tokens
Openness Index53 / 100
Input
textimagevideo
Output
text
Performance
Output speed41 tokens/s
Time to first token (TTFT)2.59 s
Time to first answer51.93 s
End-to-end response (500 t)64.3 s
Pricing
Input tokens$2.00
Cache Hit$0.250
Output tokens$6.00
Blended (3:1)$3.00
Meta-information
Tokens used (verbosity)180.0M
Evaluation cost$3172.37