Unverified50% confidenceFactExact time

早期LLaMA-2仅支持4K token上下文，而DeepSeek-V3、Qwen2.5等新一代模型已将上下文扩展至128K甚至更长

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/6/2026

First Seen

Valid until: 10/4/2026

Sources

Ollama入门：本地免费部署大语言模型的利器

bilibiliMia搞LLM7/1/2026

Related Claims

Unverified3B～4B级别模型如Qwen2.5 3B或Llama 3.2 3B在Q4量化下仅需2～2.5GB显存70% similar UnverifiedQwen3 在训练时支持长达 128K tokens 的上下文69% similar UnverifiedQwen 3.6 supports 256K ultra-long context.69% similar UnverifiedQwen3.6 27B参数版本支持最长128K的上下文窗口68% similar VerifiedQwen系列模型支持8K到128K不等的上下文窗口68% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/118894

API

curl https://kongchang.com/api/v1/knowledge/claims/118894

MCP

get_claim(id=118894)