WeMM-Embedding tops MMEB-v2: its 2B model edges out 8B rivals at a quarter of the size
tomaarsen · x · 2026-09-02
WeMM-Embedding's average scores on MMEB-v2 (78 datasets):
8B/9B class
- WeMM-Embedding-9B: 80.6
- DME-Medium (closed): 78.4
- Qwen3-VL-Embedding-8B: 77.8
2B class
- WeMM-Embedding-2B: 77.9
- Qwen3-VL-Embedding-2B: 73.2
The key takeaway: WeMM's 2B model (77.9) edges out Qwen3-VL's 8B model (77.8) at roughly a quarter of the size, while the 9B tops the closed-source DME-Medium on the overall leaderboard.
Related event: Tencent Open-Sources WeMM-Embedding, Tops MMEB Benchmarks(6 posts)→
More from Models
- Claude Fable 5.1 prompting guide: 16 official tips and an agent migration checklist — xiaohu · 2026-09-02
- Gemini 3.8 Flash benchmark charts show 71% DeepSWE, closing in on Opus 5 — testingcatalog · 2026-09-02
- Gemini 3.8 Flash rolls out scoring 71% on DeepSWE at half-price promo until 2027 — testingcatalog · 2026-09-02
- Leaked Gemini 3.8 Flash benchmarks claim wins over Opus 5 on Terminal Bench 2.1 — kimmonismus · 2026-09-02
- Gemini 3.8 Flash reportedly delivers ~Opus 5 performance at much lower cost — inductionheads · 2026-09-02
- Tester Claims Fable 5.1 Excels at Long-Horizon Tasks, Finance and Consulting 'Wiped Out' — felpix_ · 2026-09-02