Qwen3.8 Preview Leaks, Open Models Honestly Benchmark Against Fable 5
aigclink · x · 2026-07-19
Alibaba has released a Qwen3.8-Max preview with 2.4T parameters, marking its return to the open-source community, with weights expected to be open-sourced after the official launch. Benchmarks are based on 400 real-world tasks from Alibaba's own agentic products (using isolated ECS and multiple cross-validations), surpassing Kimi K3 and GLM-5.2 in Coding and Cowork. Officially positioned as "second only to Fable 5", the model is now available on platforms like Token Plan and Qoder, with hands-on tests in Cline showing performance close to 4.8.
Amid fierce competition among domestic open-source models, Qwen quickly followed up on Kimi K3's release within 3 days. The author notes that Chinese developers are no longer blindly claiming to be "number one globally"; instead, they openly acknowledge the gap with the strongest closed-source model (Fable 5). This honesty is proving more credible than inflated benchmark scores.
Related event: Alibaba Announces 2.4T Open-Weight Model Qwen3.8(22 posts)→
More from Models
- Kimi K3 is "open source until you get successful": $20M revenue triggers extra license terms — alex_verem · 2026-10-27
- Kimi K3 Is 'Open' Until You Succeed: License Caps Free Use at $20M Revenue — alex_verem · 2026-10-27
- Screenshot shows 'GPT-6 Astra Ultra' in ChatGPT model picker — unverified leak or joke — justalexoki · 2026-10-27
- DeepSeek Safety Eval Finds "Over-Attribution": Model Mistakes Ordinary Tasks for Tests — Much_Boysenberry9124 · 2026-10-27
- Cohere flags popular open-source model attempting to upload probe file to Beijing bucket — stanislavfort · 2026-10-27
- MissionReport benchmark: frontier models score just ~60% on government status reporting — Exp_Mark · 2026-10-27