Skip to content

Wire Briefings Search

Fetching from the wire…

01

02

03

04

05

06

07

08

Wire Briefings Search

Entity trail

RTX 3070 Ti

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

0

Findings

1

Edges

2

Sources

1

Corpus findings

2026-05-23 / reddit-researcherQwen3.6-35B-A3B Q4 Runs 262K Context on 8GB 3070 Ti at 30+ Tokens/secA practitioner reports running Alibaba's new Qwen3.6-35B-A3B (MoE, 3B active parameters) at Q4 quantization with 262K context on an 8GB RTX 3070 Ti at 30+ tokens/sec, pushable to 512K-1M with speed degradation past 150K. Qwen3.6 is a confirmed new release succeeding Qwen3.5 under Apache 2.0, continuing Alibaba's aggressive open-source MoE push — now rivaling much larger models on consumer hardware (74 upvotes, 30 comments).

Graph relationships

BENCHMARKED AGAINST
Qwen3.6 35B-A3B
Qwen3.6-35B-A3B runs at Q4 quantization on 8GB RTX 3070 Ti.
Source finding
BENCHMARKED AGAINST
Qwen3.6 35B-A3B -> RTX 3070 Ti
Qwen3.6-35B-A3B runs at Q4 quantization on 8GB RTX 3070 Ti.
Source finding

Source trail

Graph sources

entity graphfindings textkg edgeskg entities

RTX 3070 Ti intelligence trail | MindPattern