Model
Model explorer

Skywork-13B

OPEN
Kunlun Tech · Skywork family · released Oct 30, 2023

Bilingual (Chinese/English) base LLM trained on 3.2T tokens; best-in-class among ~13B Chinese open models at release

ReasoningCodingVisionFunction callingTool useAgentic
124.9
Elo · rank #377
Parameters
13B
Active params
13B (dense)
Context
4K tokens
Architecture
Dense transformer, 52 layers (thinner/deeper than Llama-2-13B), 4608 hidden dim
License
Skywork Community License
Languages
2+
API price (in/out)
No hosted API
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
ARC-ChallengeReasoning50.2%#125
best: Llama 3.1 405B · 96.9%
ARC-EasyReasoning78.5%#29
best: Phi-3-medium (14B) · 97.7%
C-EvalKnowledge60.6%#46
best: Qwen3.6-Plus · 93.3%
CMMLUKnowledge61.8%#25
best: Doubao-1.5-Pro · 90.9%
GSM8KMath55.8%#123
best: Llama 3.1 405B · 96.8%
HellaSwagReasoning77.4%#95
best: Claude 3 Opus · 95.4%
MMLUKnowledge62.1%#197
best: OpenAI o3 · 92.9%
PIQAReasoning79.9%#46
best: GPT-4o mini · 93.1%
TriviaQAKnowledge54.0%#38
best: Sarvam-1 (2B) · 90.6%
WinoGrandeReasoning72.2%#87
best: PaLM 2 · 90.9%
Run it locally
VRAM @ Q4
8.4 GB
VRAM @ FP16
26 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 Q4_K_M
Skywork family
Elo progression across releases
Skywork-13BOct 2023124.9
Resources
HFSkywork/Skywork-13B-baseGHSkyworkAI/Skywork
API price weights · each benchmark row carries its own source badge (see methodology)