← All features

Per-chat parameters

Every chat has a slide-out Parameters panel on the right of the window. It exposes the knobs the provider offers and remembers per-chat overrides so you can dial each conversation independently.

Simple and Advanced modes

The panel has two views:

• Simple — max tokens, num_ctx, seed, the Thinking toggle, the Low VRAM toggle, and a per-chat system prompt textarea. Deliberately terse so the common knobs are reachable without scrolling.
• Advanced — adds temperature, top_p, top_k, min_p, repeat_penalty, frequency and presence penalties, GPU layer count, and everything else the provider exposes.

Switch between them at the top of the panel. Values you set in Simple stay set when you flip to Advanced and vice versa.

The same panel is reused by Private Chat with the same Simple / Advanced split — it just closes and wipes along with the overlay.

How the values cascade

What a chat actually sends to the model is a merge of several layers. From broadest to narrowest, with later layers winning:

1. App defaults — a sensible universal fallback (for example, temperature 0.7, top_p 0.95).
2. Model defaults — parsed from the Ollama Modelfile’s PARAMETER block. Cached after the first chat with that model so subsequent loads are instant.
3. Per-model preferences — e.g. the Thinking preference you set in the Models editor.
4. Space defaults — if the chat belongs to a Space with its own pinned parameters.
5. Per-chat overrides — what the sliders in this panel save.
6. Global app overrides — for example the Settings → Features Low VRAM pin, which forces low_vram: true on every Ollama request and beats the per-chat toggle.

A Reset to defaults button in the panel header clears the per-chat overrides and falls back to the merged defaults.

Settings reference

A short description of every knob the panel exposes, grouped by what it affects. Anything tagged Ignored by OpenAI providers is sent to Ollama only — OpenAI-compatible endpoints decide that knob server-side.

Thinking

• Thinking — lets the model reason step-by-step before replying. Adds latency on long answers but often improves quality on complex prompts. Only takes effect on models that advertise thinking capability; ignored by OpenAI providers (their chain-of-thought is internal).

Sampling

• Temperature 0–1 — controls randomness. Lower stays focused and predictable; higher gets more creative and varied. Capped at 1 because output usually breaks down beyond that.
• Top-P 0–1 — nucleus sampling. The model picks from the smallest set of tokens whose probabilities sum to this value. Lower = more on-track; higher = more variety.
• Top-K 0–200 — only consider the K most likely tokens at each step. Lower is more focused; 0 disables the cap. Ignored by OpenAI providers.
• Min-P 0–0.5 — drops any token whose probability is below this fraction of the top token’s. A robust alternative to tuning Top-P/Top-K. Ignored by OpenAI providers.

Length

• Context Length (num_ctx) — how much conversation history the model can see at once. Larger windows remember more but use more VRAM. Snaps to power-of-two stops (4K → 1M). Ignored by OpenAI providers — the server decides.
• Max Tokens 128–32768 — hard cap on the length of a single reply. Generation stops here even if the model isn’t finished.

Repetition

• Repeat Penalty 0.8–2.0 — penalizes tokens the model has just used, discouraging loops. 1.0 is off; much above 1.3 starts to sound robotic. Ignored by OpenAI providers.
• Frequency Penalty −2 to 2 — pushes down tokens in proportion to how often they’ve already appeared in this reply. Negative values encourage repetition instead.
• Presence Penalty −2 to 2 — flat one-time penalty for any token that has already appeared. Nudges the model toward fresh vocabulary and new topics.

Performance

• GPU Layers (num_gpu) — how many model layers to offload to the GPU. Blank = Ollama auto-detects; 0 = CPU only; a positive integer = explicit layer count (useful when a model almost fits but its KV cache spills off the GPU). Ignored by OpenAI providers.
• Low VRAM — trades throughput for footprint (smaller batches, leaner KV cache). Flip on when you’re hitting OOMs and num_gpu alone isn’t enough. Can be pinned on globally in Settings → Features, which forces the per-chat toggle on. Ignored by OpenAI providers.

Reproducibility

• Seed — a fixed integer makes the model’s output reproducible for the same input. Leave empty for a fresh random seed each run. The dice button generates a random seed in place.

Prompt layering

• Persona — defines the assistant’s role (e.g. Code Reviewer, Writing Editor). Layered into the system prompt at send time. See Personas and tones.
• Tone — style modifier appended after the system prompt. Falls back to the global default (Settings → General) when not set per chat.
• Additional instructions (this chat) — free-form per-chat system prompt text. Sits between the persona and the tone. Empty falls back to the global custom instructions.

Thinking toggle

Visible for Ollama models whose capabilities include thinking. It sets the think parameter on the chat request and is ignored by OpenAI-compatible providers. The default for new chats comes from Settings → Features → Thinking.

Low VRAM toggle

Ollama-only. Forces smaller batches and a leaner KV cache — useful when you are tight on GPU memory. The per-chat toggle is overridden if Settings → Features → Low VRAM mode is enabled globally; in that case the panel shows the toggle pinned and disabled with a pointer back to the global setting. See Low VRAM mode.

Per-chat system prompt

The textarea at the bottom of the Simple view sets a system prompt for this chat only. It overrides any global custom instructions and Space-level instructions for the chat it lives in.