priorsBuilder

Alibaba · released 2026-08-12

Qwen 3.8

Measured on 5 benchmarks.

General frontier

Not ranked in this mix — too little measured data for its benchmarks.

Your mix’s benchmarks

0 of 8 measured

This model has no results on your mix’s benchmarks.

All 5 benchmarks

show 5 more

coding

WeirdML v2

127 results
  1. #26default of 12775.2%

Leader: GPT-6 Astra Pro · 93.6%

reliability

Tool-call error rate

172 results
  1. #164default of 1729.5%

Leader: Gemini 4 · 0.0%

reliability

Uptime (7d)

182 results
  1. #123default of 18298.8%

Leader: GPT-5 Pro · 100.0%

speed

Provider latency

178 results
  1. #111default of 1781.6s

Leader: Qwen 2.5 7B · 0.2s

speed

Provider throughput

176 results
  1. #46default of 17685.0

Leader: Muse Spark 1.1 · 204.5