Kimi K3, GLM-5.2 and Qwen 3.8 battle for the open-weight crown
johnseach · x · 2026-07-27
A July 2026 showdown compares Kimi K3, GLM-5.2, and Qwen 3.8 as the new open-weight contenders.
- Kimi K3 (Moonshot): about 2.8T parameters, extreme sparsity, 1M context, native vision, and claimed top capability among open models; weights are said to be dropping today.
- GLM-5.2 (Zhipu): about 753B total / 40B active, 1M solid context, sparse attention with IndexShare, fully open under MIT, and positioned as the best value for pure coding and long-horizon agents.
- Qwen 3.8 (Alibaba): claimed 2.4T parameters, strong multimodal support, around 1M context, still in Max-Preview, with open weights promised soon.
The author’s take: Kimi K3 raises the ceiling, GLM-5.2 is the most practical and cheapest, and Qwen 3.8 is the big multimodal wildcard still waiting to fully land.
More from Models
- Qwen3.6-27B speculative decoding speeds up as quantization gets heavier — thavoc77 · 2026-07-27
- Kimi Delta Attention cuts KV cache by 75% and speeds million-token decoding by 6× — johnseach · 2026-07-27
- Leak: Moonshot to Release Kimi K3 with 2.8T Parameters and 1M Context — johnseach · 2026-07-27
- Arena screenshot says Gemini 3.5 Pro beats Opus 5 in two tests — Last_Conclusion_8984 · 2026-07-27
- Binance Android app reportedly disappears from Google Play in parts of Europe — RSync25 · 2026-07-27
- Kimi K3 weights go public, and the jokes start immediately — uwukko · 2026-07-27