Alex Getmanearly AI signals
EN
Alex Getman @alexgetmancom · · Updated

Releases

QWEN3.8 FLASH NEXT IS OUT

🚨 QWEN3.8 FLASH NEXT IS OUT

The leaked specs were real:

  • 125B multimodal MoE
  • Only 6B active
  • 262K native context, up to 1M

Qwen says it beats Claude Opus 4.6 across coding, agentic tasks, tool use and several multimodal benchmarks.

Table comparing models: Qwen3.8-Flash-Next (125B params, 6B active) shows top scores in coding benchmarks (DeepSWE 1.1, SWE-bench Pro, SWE-bench Multilingual, NL2Repo-Bench) versus Qwen3.8-27B, Qwen3.7-Plus, DeepSeek-V4-Flash-0731, and Claude-Opus-4.6 (Max).
Table comparing models: Qwen3.8-Flash-Next leads in agent tasks (CoWorkBench, JobBench, Agents' Last Exam, Toolathon Verified) and general benchmarks (IFBench, GPQA Diamond, HLE, LiveCodeBench v6) compared to Qwen3.8-27B, Qwen3.7-Plus, DeepSeek-V4-Flash-0731, and Claude-Opus-4.6 (Max).
ClaudeAlibabaQwen
Русская версияMarkdown

Related posts