Modelpublic

RobiLabs/Vera

sourceHugging Facemitupdated 2mo agoView on Hugging Face

1likes62downloads

Model Card

Vera

Vera is a 30B-A3B Mixture-of-Experts (MoE) causal language model developed by Robi Labs. Designed for high performance and lightweight deployment, Vera offers deep conversational capability, software engineering skills, and reasoning support.

Model Summary

—Developed by: Robi Labs
—Model Type: Mixture-of-Experts (MoE) Causal LLM (VeraMoeLiteForCausalLM)
—Languages: English (en)
—License: MIT
—Context Length: 200k tokens

Performances on Benchmarks

Benchmark	Vera-30B-A3B-MoE	Qwen3-30B-A3B-Thinking-2507	GPT-OSS-20B
AIME 25	91.6	85.0	91.7
GPQA	75.2	73.4	71.5
LCB v6	64.0	66.0	61.0
HLE	14.4	9.8	10.9
SWE-bench Verified	59.2	22.0	34.0
τ²-Bench	79.5	49.0	47.7
BrowseComp	42.8	2.29	28.3

Key Features

—Architectural Efficiency: Built on a custom MoE architecture (veramoelite) balancing fast inference times with expert layer routing.
—Extended Context: Supports up to 200,000 tokens context window for extensive reasoning, document understanding, and multi-turn conversations.
—Task Capability: Highly optimized for general reasoning, mathematical evaluation, and multi-turn chat alignment.