Hey All,
We just wrapped a hands-on round with our EmailEval framework: here’s what I discussed in the video and my current top open-source picks for email-generation–focused models on Hugging Face:
Top Open Source Email Models – BrainDrive
Top 3 (with quick stats)
Tulu-2-7B (AI2) — allenai/tulu-2-7b · Hugging Face
Score 8.89 | 6-metric profile (Clarity, Length, Spam Risk, Personalization, Tone, Hygiene) | 7B params | OpenRAIL license | EN
StarChat-Beta (Hugging Face H4) — HuggingFaceH4/starchat-beta · Hugging Face
Score 8.54 | 6-metric profile | ~7B params | Apache-2.0 | Multilingual (EN strong)
LFM2-1.2B (Liquid AI) —LiquidAI/LFM2-1.2B· Hugging Face
Score 8.44 | 6-metric profile | 1.2B params | Optimized for efficiency | EN
How we ranked them (EmailEval by BrainDrive)
EmailEval is our evaluation workflow for AI-generated emails. We score models on 6 practical metrics—from Clarity, Personalization, length, and Spam Score to Tone and Grammatical Hygiene. Individual scores roll up into a weighted total, which determines ranking.
Docs (scoring & math):
ModelMatch/EmailEval/Docs · GitHub
Workflow we used
Model shortlist (20+ HF candidates) →Multi prompts → Responses per model → EmailEval scoring → Weighted aggregation → Ranking.
Try it yourself
Code toolkit: ModelMatch/EmailEval · GitHub
No-code evaluator: BrainDrive/EmailEval – Hugging Face Space
About ModelMatch
ModelMatch helps you discover the most suitable open-source model for your domain and task—starting with summarization, expanding into therapy, and now email generation.
If you test other models or get different results, ping us; happy to compare notes.
Regards,
Navaneeth