Revision History
Best Open-Source LLMs · 4 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: full refresh as of October 30, 2026. Rankings now use the Artificial Analysis Intelligence Index v5.0 instead of the July v4.1 scores; picks updated to DeepSeek V4.1 Flash 1017, GLM-5.3, MiMo-V2.6-Pro, MiniMax-M3.1-Flash, Muse Spark Open, Mistral Large 4, Qwen3.8 27B and Nemotron 3.5 Super VL; GLM-5.2's release date corrected to June 16; the unsupported Kimi K2.6 score removed; license terms and hardware needs now quoted and computed from the developers' own repositories.
Version 4 · Oct 30, 2026, 11:13 AM
- Correction: GLM-5.2's size changed from ~753B to Z.ai's 744B-A40B label, with the Hugging Face count explained; corrected the hardware note (the FP8 weights, about 756 GB, exceed an 8x H100 node, and 4-bit GGUF builds are 365 to 467 GB); fixed a broken GPT-5.5 link.
Version 3 · Oct 30, 2026, 09:46 AM
- Added 6 contextual internal links
Version 2 · Jul 23, 2026, 04:06 PM