# Cohere Chat Canonical page: https://fuser.studio/models/cohere-chat Cohere Chat in Fuser runs command-a, command-r and command-r7b as text nodes. Set a system prompt and chain the output into image, video or audio steps. ## Specs | Spec | Value | | --- | --- | | Versions | command-a, command-r7b, command-r, command-r-plus | | Inputs | Prompt (text), System Prompt (text) | | Output | text | Fluent copy in 23 languages, long briefs digested in one pass, and a model tier for every speed and budget Creator: Cohere [Draft the copy](https://app.fuser.studio) ## How Cohere Chat works Choose one of four Command models, brief it through a system prompt, then iterate with a few dials. The big lever is the system prompt, which decides whether you get clean copy or a chatty reply. ### Pick your model tier Choose command-a for long briefs, command-r for everyday drafting, or command-r7b for fast bulk variations. ### Set voice and paste source Put role, tone and format in the system prompt, such as no Markdown and no preamble, then paste your brief or script into the prompt. ### Tune and iterate Lower temperature for localization, raise it for taglines, and hold the seed fixed while you refine your wording. ## What Cohere Chat is good at Multilingual fluency, long-document appetite and a speed-to-quality ladder. This is a text-in, text-out collaborator for copy, scripts and prompt drafting, not a visual critic. ### Multilingual campaign copy Coverage spans 23 languages, and the output keeps rhythm and imagery instead of translating word for word. Use command-r at low temperature for routine localization, or step up to command-a when the instructions get complex. ### Long-brief digestion Paste a whole brand book or transcript as text and get the essentials back in a single pass. command-a offers a 256K context for the heaviest briefs; command-r and command-r-plus work with 128K. ### Character dialogue that holds command-r-plus keeps character and lore steady across long contexts, which suits scene drafts, roleplay and dialogue passes. Tell it in the system prompt not to resolve conflict quickly to counter its upbeat bias. ### Bulk variations at speed Ask for a numbered list and it returns a batch of variations in one go, from exhibition titles to wall-label openers. Use command-r7b when speed matters most and Command A when every line has to hold up; a stop sequence trims each to a clean item. ## Made with Cohere Chat A gallery of written outputs: localized taglines, a tense scene, a shot list, a prompt for a video model and a song brief. Each shows the exact instruction that produced it. ### Campaign copy: one product, three languages ### Scriptwriting: dialogue that refuses to soften ### Storyboard: a text scene broken into shots ### Prompt writing for the next node on the canvas ### Song brief: sonic direction plus a chorus ## What people build with Cohere Chat Copywriters localizing campaigns, art directors digesting briefs, writers drafting dialogue and creators feeding prompts downstream all reach for it when words, languages or volume are the bottleneck. ### Global campaign localization Brand writers adapt launch copy for several markets in one pass, keeping rhythm and sensory imagery across languages. Run it at low temperature and ask for labeled versions only. ### Art direction from long briefs Art directors and producers paste brand books, transcripts and lore bibles as text and get a tight brief back. command-a takes the longest sources and the most complex instructions. ### Dialogue and roleplay drafting Writers and roleplayers use command-r-plus to hold character and lore across long exchanges. A clear system prompt about tone keeps scenes from resolving too neatly. ### Prompt writing at volume Creators feeding image, video and audio nodes use command-r7b to spin out many prompt variants quickly. Ask for a specific format, like one paragraph or comma-separated tags. ### Shot lists from written scenes Directors and short-form creators turn a scene on the page into a shot list, with sizes, moves and on-screen text, ready to carry into a video workflow. ## Cohere Chat questions ### What's the difference between the default route and the alternative route? The default route is the main way to reach all four Cohere models here, and it suits everyday drafting, localization and long-brief digestion. The alternative route reaches the same chat models when the default is unavailable, or when you want to compare the same prompt on another access path. Parameter support and defaults may differ on it, so keep settings simple: prompt, system prompt and temperature. Cost is listed as 0 on both, so neither gives a price signal. ### Which Cohere Chat model should I use for my task? Match the tier to the job: - **command-a**: the flagship with a 256K context, best for long briefs and complex instructions. - **command-r-plus**: 104B, strong long-form prose and grounded rewriting. - **command-r**: the default, balanced for everyday drafting. - **command-r7b**: 7B, fastest and cheapest, good for bulk taglines and prompt variations. It is weaker on long scripts that must stay consistent. ### Can Cohere Chat read or critique images and video? No, these models are text-focused, so don't rely on them for image or video analysis. Layout, color and typography critique from an attached image is a poor fit. Treat attachments as text sources such as scripts, notes and briefs, and paste source material straight into the prompt. ### How do I stop it adding Markdown, intros and follow-up offers? Say so in the system prompt, for example: "No Markdown, no preamble, return only the copy." Vague style notes like "write casually" tend to get overridden by its polished default. command-a is chatty by default, while command-r7b is more compact. ### Why do its stories resolve conflicts so fast, and how do I fix it? The models lean upbeat and reach happy resolutions quickly, especially in fiction. In the system prompt, state the tone and say conflicts should not resolve quickly. Phrases like "keep the conflict unresolved" and "short, clipped sentences" help. Without heavy steering, unflinchingly dark or ambiguous endings are a weak spot. ### Who makes Cohere Chat and when were the models released? The models are made by Cohere. The versions in this node were released between 2024-08-30 and 2025-03-13, spanning command-r and command-r-plus (08-2024), command-r7b (12-2024) and command-a (03-2025). ## Try Cohere Chat on Fuser Fluent copy in 23 languages, long briefs digested in one pass, and a model tier for every speed and budget [Draft the copy](https://app.fuser.studio) ## Related models - [OpenRouter Chat](https://fuser.studio/models/openrouter-chat.md) - [Perplexity Chat](https://fuser.studio/models/perplexity-chat.md) - [ChatGPT](https://fuser.studio/models/chatgpt.md) - [Google Gemini Chat](https://fuser.studio/models/google-gemini-chat.md) - [Groq Chat](https://fuser.studio/models/groq-chat.md) - [Mistral Chat](https://fuser.studio/models/mistral-chat.md) Node reference: [Cohere Chat docs](https://docs.fuser.studio/docs/nodes/llm/cohere-chat.md)