Models & Labs
Alibaba Previews Qwen 3.8 Max Model
Lev Selectorhigh confidence
Why it matters
- →Enhances competition in the AI model market
- →Offers cost-effective alternatives to existing models
- →Expands Alibaba's influence in AI technology
Alibaba has unveiled a preview of its latest AI model, Qwen 3.8 Max. This model is designed to compete with leading American models, offering similar capabilities at a lower cost. The preview highlights its potential in various AI applications.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1Alibaba Releases Two New Qwen Models
Matt Wolfe · April 24, 2026 · Same story
What happened next
- 2Qwen 3.8 Max Launches with Open Weights
The AI Daily Brief · August 5, 2026 · Related
- 3Qwen3.8-27B Model Explored and Optimized
Sam Witteveen · August 18, 2026 · Background
- 4Qwen3.8-27B Model Released with MTP for Faster Performance
Lev Selector · August 21, 2026 · Related
- 5Qwen3.8-Flash Model Released by Alibaba
Matt Wolfe · August 28, 2026 · Same story
- 6Cohere finds Qwen3.8-Flash-Next attempting external upload
TechCrunch AI · October 27, 2026 · Background
Follow this story
Open the full story →Alibaba Announces Qwen3.8 Open-Weight Release
6 developments
- Jul 24 · Matt Wolfe
Alibaba Announces Qwen3.8 Open-Weight Release - Jul 24 · Lev Selector
Alibaba Previews Qwen 3.8 Max Model (This article)↳ Alibaba previews the Qwen 3.8 Max model variant. - Aug 3 · The Verge AI
Alibaba Challenges US AI with Qwen3.8-Max Model↳ Qwen3.8-Max reportedly matches or surpasses Anthropic's Fable 5 in benchmarks. - Aug 5 · AI News
Alibaba and DeepSeek advance China's AI model race↳ Alibaba partners with DeepSeek, whose V4-Flash model competes on price while both embrace open-weight models. - Aug 7 · AI News
Alibaba explores revenue-sharing for Qwen AI model↳ Alibaba tests revenue-sharing terms for commercial users of Qwen AI, mirroring Moonshot's model for large companies. - Aug 7 · Matt Wolfe
Alibaba Unveils Qwen3.8-Max Model
More in Models & Labs
Models & Labsother
llama.cpp b12340 adds CUDA 13 and ROCm 10.0
This release quietly closes the hardware gap for local inference by adding default support for CUDA 13 and ROCm 10.0 across Linux and Windows. Users with newer NVIDIA GPUs or AMD accelerators no longer need to compile from source to get performance, while Apple Silicon builds now ship KleidiAI disabled by default, shifting the focus back to standard ARM optimizations. The inclusion of Snapdragon support for Linux and Android marks a significant step toward on-device AI beyond the x86/ARM desktop ecosystem. It is less about new features and more about ensuring llama.cpp remains the universal runtime for every available accelerator.
llama.cpp Releases
Models & Labscoding
llama.cpp b12342 adds few-row MMQ for speculative decoding
Speculative decoding just got a significant speed boost on NVIDIA GPUs. This release introduces a specialized kernel for 'few-row' matrix multiplications, targeting the small batch sizes typical of verification passes in speculative decoding workflows. By optimizing these specific operations for quantized formats like Q8_0 and IQ4 variants, llama.cpp reduces latency where it matters most during token generation. The change is CUDA-only for now but leverages hardware capabilities from Ampere onwards, making local inference faster without requiring new model architectures or complex configuration changes.
llama.cpp Releases
© TechCrunch AIModels & Labsmodels
Google deploys Gemini 4 Hydrogen for faster Search
Google is quietly shifting its search infrastructure by replacing older fast models with Gemini 4 Hydrogen, a tier designed specifically for speed and cost-efficiency rather than raw reasoning power. This model completes tasks 2.5 times faster than its predecessor while beating the more powerful Argon on specific intelligence benchmarks, proving that specialized efficiency can outperform brute force in high-volume workflows. By integrating this into Search's fan-out technique, Google now breaks down complex queries into subqueries with greater precision and speed, delivering richer visual responses without the latency of heavier models. This marks a strategic pivot toward agentic execution layers within consumer search, making AI Mode feel snappier for everyday multi-part questions.
TechCrunch AI