gooneyryan/hermes-webui
1
0
Fork 0
mirror of https://github.com/nesquena/hermes-webui.git synced 2026-05-27 04:00:37 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs(models): update Qwen3.6 35B replaces Qwen 3.5 27B — April 23, 2026

Browse Source
This commit is contained in:
nesquena-hermes
2026-04-23 07:11:34 +00:00
parent 7a7129b1c9
commit 94ff339dd8
1 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
models/index.html
+9 -9
View File
@@ -617,7 +617,7 @@
<div class="hero-badge">2026 Model Guide</div>
<h1>Best AI models for <em>Hermes</em></h1>
<p class="hero-sub">Top picks across coding, writing, search, and reasoning — so you know exactly what to plug in and why.</p>
<p class="hero-note">Data from SWE-bench Pro, GPQA Diamond, Chatbot Arena, and BenchLM. Updated April 21, 2026. <a href="https://lmarena.ai/leaderboard" target="_blank">Source →</a></p>
<p class="hero-note">Data from SWE-bench Pro, GPQA Diamond, Chatbot Arena, and BenchLM. Updated April 23, 2026. <a href="https://lmarena.ai/leaderboard" target="_blank">Source →</a></p>
</section>
<!-- =========================================
@@ -1216,17 +1216,17 @@
<div class="model-card" style="--model-color: #f0a500;">
<div class="rank-badge silver">🥈</div>
<div class="model-info">
<h3>Qwen 3.5 27B</h3>
<div class="model-meta">Alibaba &middot; Apache 2.0 &middot; 262K context &middot; ~17 GB VRAM (Q4)</div>
<div class="model-why">Best coding benchmark of any model that fits on a single consumer GPU — 72.4% SWE-bench Verified and 85.5% GPQA Diamond, beating cloud models twice its price. 262K native context (extensible further via YaRN) handles large codebases in a single pass. Dual-mode: fast direct answers or slow chain-of-thought reasoning when you need it. Runs on an RTX 4090 or M2/M3 Max with 24 GB VRAM. 95.0% on IFEval, which beats GPT-5-mini. Released February 2026, multimodal (text + vision), Apache 2.0. The pragmatic local pick for developers.</div>
<h3>Qwen3.6 35B</h3>
<div class="model-meta">Alibaba &middot; Apache 2.0 &middot; 128K context &middot; ~18 GB VRAM (Q4)</div>
<div class="model-why">Released April 14, 2026. 73.4% SWE-bench Verified and 86.0% GPQA Diamond — the strongest coding and reasoning scores of any open-weight model that runs on consumer hardware. Sparse 35B/3B-active MoE architecture runs on a 16 GB Mac Mini or RTX 4090. Terminal-Bench 2.0: 51.5% (up from 41.6% on Qwen3.5-27B). Multimodal (vision + language), supports thinking mode. Supersedes Qwen 3.5 27B on all major coding benchmarks.</div>
<div class="model-pills">
<span class="pill gold">SWE 72.4%</span>
<span class="pill blue">262K context</span>
<span class="pill green">24 GB VRAM</span>
<span class="pill gold">SWE 73.4%</span>
<span class="pill blue">128K context</span>
<span class="pill green">16 GB VRAM</span>
</div>
</div>
<div class="model-score">
<span class="score-val">72.4%</span>
<span class="score-val">73.4%</span>
<span class="score-label">SWE-bench</span>
</div>
</div>
@@ -1384,7 +1384,7 @@
</div>
<div class="picker-card">
<div class="use-case"><span class="uc-icon">🖥️</span> Run privately</div>
<div class="recommendation">Gemma 4 31B or Qwen 3.5 27B</div>
<div class="recommendation">Gemma 4 31B or Qwen3.6 35B</div>
<div class="why-short">No API key, no data leaving your machine. Best two on consumer hardware.</div>
</div>
</div>