Author: Hutch

Sillytavern Preset Comparison Tool: Introducing PresetLab

Screenshot of PresetLab interface showing scattered presets and controls.
~ Hutch ~ Leave a comment

Sillytavern Preset Comparison Tool: Introducing PresetLab

The Curse of Knowing Just Enough

SillyTavern presets. The lifeline. Unfortunately, there’s really no clean way to “do” a SillyTavern preset comparison. It’s made more difficult by the fact that you can’t really see output side-by-side, so you’d have to swap around and mess around just to see the output a preset can give you. On top of that, the good community presets are often discussed, but that leaves little spotlight for the other smaller presets the community worked hard on. Those get little to no discussion.

With no quick, braindead way to see the contents of a preset, no way to run the same prompt through two presets at once, no central list for presets, I decided to try something out. So I made PresetLab.

Continue reading →

SillyTavern Weekly: DeepLore v2.5, Natural Extended v0.9, GLM 5.2 Settles In — 6/16 – 6/22

~ Hutch ~ Leave a comment

SillyTavern weekly news for the week of June 16 to June 22, 2026. The model story this week is still GLM 5.2 *yawn* (kidding, I’m just salty from z.AI’s shift). Putting that to the side, the real story is the user-side tooling cluster: DeepLore v2.5, Natural Extended v0.9, Leonardo, Zebede, and a quiet but real character-card workflow shift. Megumin V8. GitHub quiet. Also, what model is Owl Alpha on OR?.. That’s not part of the news. That’s my personal question. Stealth models drive me nuts.

Continue reading →

GLM 5.2 Presets for SillyTavern

~ Hutch ~ Leave a comment

GLM 5.2 Presets for SillyTavern

On a day such as this, one could not have expected better news for the GLM faithful. GLM 5.2 just dropped on OpenRouter the second week of June, and the early community response has been strong. GLM 5.1 was already ranked #2 overall by the Freaky Frankenstein creator (per their tier list, output 9/10, censorship: zero), and 5.2 is just more of that on more headroom. The chain-of-thought architecture is the real story though. Give it a tiered preset and it processes each layer before it spits out a response, like a DM checking session notes. That combination is what makes GLM feel less like a model and more like a partner.

Continue reading →

DeepSeek V4 Flash Presets for SillyTavern

~ Hutch ~ Leave a comment

DeepSeek V4 Flash Presets for SillyTavern

DeepSeek V4 Flash had a rocky launch, and I am not going to sugarcoat that. Community reports cited POV confusion, stereotypical RP output, and weak prompt adherence. The big wrinkle is that V4 ships with built-in RP instructions that actively fight custom presets, which is why it needs special handling. The good news: V4 Flash is “surprisingly good” per the community as of June 2026, and DeepSeek has committed to ongoing roleplay-specific tuning based on community feedback. The 1M context window and permanent 75% price reduction make it one of the most accessible frontier models for marathon sessions. Use V3.2 when you need reliability, V4 Flash when you need 1M context or cost.

Continue reading →

DeepSeek V3.2 Presets for SillyTavern

~ Hutch ~ Leave a comment

DeepSeek V3.2 Presets for SillyTavern

I tried V3.2 again last month after bouncing off it earlier in the year, and I am not going to lie, it caught me off guard. “Apparently all this model needed was a quality preset. My goodness can it cook” is the vibe in the megathreads right now, and I’m inclined to agree. The community has it ranked #3 overall (per the Freaky Frankenstein tier list as of June 2026), the FF creator uses it interchangeably with GLM 5.1 depending on scene type, and the prose is vivid in a way the front-page OpenRouter rankings don’t really capture. The one real limitation is the 163K context window, which is lower than the 1M-token models, so very long sessions can fall off. That’s a tradeoff, not a dealbreaker.

Continue reading →

Claude Opus 4.6 Presets for SillyTavern

~ Hutch ~ Leave a comment

Claude Opus 4.6 Presets for SillyTavern

Look. I use Claude (sparingly). I have used it for a long time (sparingly!), and I am not going to sit here and pretend the bill doesn’t hurt. The prose is still the best in the business, the emotional range is something the other models can’t touch, and the way it handles character dynamics without you having to spell them out is a genuine superpower. The friction, beyond the price, is that Anthropic has been quietly adding new API-level safeguards to older Claude models, so things that worked last month may suddenly refuse. That part is annoying. It is what it is.

Continue reading →

Gemini 3.5 Flash Presets for SillyTavern

~ Hutch ~ Leave a comment

Gemini 3.5 Flash Presets for SillyTavern

Gemini 3.5 Flash presets are plentiful. If you’ve used Gemini 3.5 Flash for any length of time, you already know the deal. It’s fast, it’s cheap, and the prose is the best the Gemini line has produced to date. “Less ‘Geminisms'” is the consensus, and that tracks with what I’ve seen in long sessions. The main practical tip is to turn off streaming on AI Studio, otherwise the API-level filter will block your message sends mid-stream. On the censorship question, Flash is less restrictive than previous Gemini models for NSFW, but violence has some filter friction on AI Studio specifically. NanoGPT and OpenRouter tend to be cleaner. The 1M context window and Flash-tier pricing make it a strong value pick.

Continue reading →

SillyTavern Weekly: GLM 5.2, DeepSeek v4, Fable 5 — 6/8 – 6/15

~ Hutch ~ Leave a comment

SillyTavern weekly news. We’re going to ignore that there was no SillyTavern weekly news for the entire month of May. Or any post, really (Did I say I was sorry?)

The SillyTavern weekly news of June 8 to 15, 2026 was a model release week, which is also the only kind of week that matters around here. *GLM 5.2* finally hit OpenRouter, a few days late but no less hyped for the wait. *DeepSeek v4* turned out to be, against all odds, actually good. *Fable 5* launched and immediately started a culture war. Anthropic put new safeguards on old models in the kind of move that makes you check the date twice. And underneath it all, the SillyTavern GitHub ship sat quiet for the sixth week running.

Let’s get into it.

Continue reading →

I’m Still Alive, Just Buried in Tabs

~ Hutch ~ Leave a comment

The Curse of Juggling Too Many Things at Once

It’s been a minute. Or two. Or technically closer to a couple of months, if we’re counting the calendar instead of pretending (and I’m good at pretending. That’s what RP is all about.)

I didn’t disappear on purpose. I just kept picking up one more thing, then one more thing after that, and somewhere between “I’ll write the post next weekend” and now, the next weekend stopped coming back around. So. Here we are. I’m back, briefly, to admit that I went dark and to promise there’s actually a reason for it. GLM vs MiniMax for RP

Continue reading →

SillyTavern Weekly: DeepSeek V4, Kimi K2.6, GLM-5.1 — 4/27/26

~ Hutch ~ 2 Comments

SillyTavern weekly news: Seven days of morning briefings compiled so you don’t have to scroll Reddit at 4am. You’re welcome.

The SillyTavern weekly news of April 20 to 27, 2026 was a big one if you live and breathe model releases and community drama. DeepSeek V4 dropped with a flashier little brother. Z.AI kept everyone in a permanent state of confusion about whether their legacy coding plans were about to get gutted. Kimi K2.6 launched and divided the room like every Kimi release does. And underneath all the noise, the SillyTavern GitHub ship stayed steady with critical bug fixes and a shiny new model on the approved list.

Let’s get into it.

Continue reading →