Model
Model explorer

StableLM Zephyr 3B

OPEN
Stability AI · StableLM family · released Dec 7, 2023

DPO-tuned chat model (UltraFeedback-style mix) on StableLM-3B-4E1T; capable assistant aimed at edge-device deployment.

ReasoningCodingVisionFunction callingTool useAgentic
-0.4
Elo · rank #403
Parameters
2.8B
Active params
2.8B (dense)
Context
4K tokens
Architecture
Dense decoder-only transformer, DPO-tuned from StableLM-3B-4E1T
License
Stability AI Community License (non-commercial by default; commercial use requires Stability AI membership)
Languages
1+
API price (in/out)
No hosted API
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
AGIEvalReasoning33.2%#35
best: OLMo 3-Think 32B · 88.2%
AlpacaEvalHuman preference76.0%#6
best: Tülu 2+DPO 70B · 95.1%
ARC-ChallengeReasoning46.1%#132
best: Llama 3.1 405B · 96.9%
BIG-benchReasoning35.3%#6
best: Chinchilla · 65.1%
GSM8KMath42.1%#142
best: Llama 3.1 405B · 96.8%
HellaSwagReasoning74.2%#116
best: Claude 3 Opus · 95.4%
MMLUKnowledge46.2%#242
best: OpenAI o3 · 92.9%
MT-BenchHuman preference6.64#59
best: Hunyuan-Large (A52B) · 9.4
TruthfulQAKnowledge46.5%#61
best: Phi-3.5-MoE (16x3.8B, 6.6B active) · 77.5%
WinoGrandeReasoning65.5%#118
best: PaLM 2 · 90.9%
Run it locally
VRAM @ Q4
2.2 GB
VRAM @ FP16
5.6 GB
Fits on (Q4)
RTX 3060 12GBRTX 4070 Ti 16GBRTX 3090 24GBRTX 4090 24GBRTX 5090 32GBM4 Pro 48GBM3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB
Throughput data unavailable.
Quantizations
GGUF · GPTQ
StableLM family
Elo progression across releases
StableLM-Alpha 7BApr 2023-552.2StableLM-3B-4e1tOct 2023-203.6StableLM Zephyr 3BDec 2023-0.4
Resources
HFstabilityai/stablelm-zephyr-3bGHStability-AI/StableLM
API price weights · each benchmark row carries its own source badge (see methodology)