\u003C/svg>`,\n close: `\u003Csvg xmlns=\"http://www.w3.org/2000/svg\" fill=\"none\" viewBox=\"0 0 24 24\" stroke-width=\"1.5\" stroke=\"currentColor\" aria-hidden=\"true\" style=\"height: 30px; color: white;\">\u003Cpath stroke-linecap=\"round\" stroke-linejoin=\"round\" d=\"M6 18L18 6M6 6l12 12\">\u003C/path>\u003C/svg>`\n },\n icon: {\n /* path to a remote image for the widget button, takes precedence over the svg.widget property */\n widgetIcon: \"https://cdn.example.com/widget-icon.png\",\n }\n};\n\u003C/script>\n```\n\n## Other settings\nIn addition to overriding the default configuration shown above, you can also configure other settings via our **window.chatThingConfig** global.\n\n```typescript\nexport type TLocale = \"de\" | \"es\" | \"fr\" | \"it\" | \"jp\" | \"cn\" | \"el\";\n\nexport interface IChatThingConfig {\n /** locale for the chat interface, we currently support English, Spanish, French, German, Japanese, Chinese & Greek */\n locale?: TLocale;\n /** an alternative to setting the **id** attribute on the chat widget script tag */\n botId?: string;\n /** an initial question which will be sent to your bot when the chat widget is opened */\n initialQuestion?: string;\n /** will cause the chat window to automatically show after N ms (1000ms = 1second) */\n autoOpenDelay?: number;\n /** css overrides for all the interface elements we add to your site, see above for our defaults */\n css?: {\n widget?: string;\n window?: string;\n iframe?: string;\n messagePreviewContainer?: string;\n messagePreviewBox?: string;\n messagePreviewText?: string;\n messagePreviewCloseButton?: string;\n messagePreviewCloseIcon?: string;\n messagePreviewIconWrapper?: string;\n messagePreviewIconImg?: string;\n customAssistantIcon?: string;\n bubbleCustomAssistantIcon?: string;\n };\n /** override our default icons, these should be provided as inline svg content */\n svg?: {\n widget?: string;\n close?: string;\n };\n /** css colour values for interface elements, your css overrides will take precedence over these */\n colour?: {\n primaryColour?: string;\n widgetColour?: string;\n widgetColourInverted?: string;\n };\n /** provide a path to an asset for the widget button, this takes precedence over the svg icon */\n icon?: {\n widgetIcon?: string;\n };\n /** the message shown in a chat bubble above the widget button */\n firstMessage?: string;\n /** the time in seconds to wait before showing the first message in a chat bubble */\n messagePreviewDelay?: number;\n /** theme applied to the chat interface, this isn't always relevant if you choose override our default css */\n theme?: \"dark\" | \"light\";\n url?: string;\n}\n\n```","2023-08-01T14:38:10.129Z","2024-05-08T15:18:49.687Z","2024-02-02T12:27:33.648Z",null,{"data":224},{"id":20,"attributes":225},{"title":42,"content":226,"slug":43,"position":8,"createdAt":227,"updatedAt":228,"publishedAt":229,"locale":208,"long_title":222},"Your bot advanced settings let you tweak and adjust your bot and the underlying AI model that powers it. \n\n---\n\n## Where are my bot advanced settings?\n\nTo access your bot advanced settings, go to the bot dashboard, navigate to the “General settings” tab, and scroll to the “Model settings (advanced)” area beneath the general settings.\n\n![advanced settings](https://res.cloudinary.com/djyjvrw5u/image/upload/v1721379183/advanced_settings_b9fd792501.png)\n\n## Choosing the bot model (GTP-3/GPT-4)\n\n{% callout title=\"Only available on some plans\" type=\"warning\" %}\nModel selection is only available to Standard or above subscribers.\n{% /callout %}\n\nYou can switch the underlying AI model your bot uses by using the model drop down in the advanced settings section. \n\nFor more information about the available AI models read here: [Available models](https://chatthing.ai/docs/available-models)\n\n## Adjusting your bots creativity (temperature)\n\nUse the creativity slider to adjust how creative your bot is when answering questions. Use a low value if you want your bot to:\n\n- Answer the same question the same way each time\n- Be less likely to make things up\n- Be more likely to quote your data sources word for word\n- Be more likely to follow your prompt strictly\n\nUse a high value if you want your bot to be more creative and do the following:\n\n- Answer the same question with more variations each time it is asked\n- Fill in gaps in the knowledge provided by your data sources\n- Be more likely to make things up\n\n## Adjusting the document relevance slider\n\nWhen you ask your bot a question, we load up relevant parts of your data sources for the bot to use within its answers. By default, we only give the bot data we think is very relevant to your query, but adjusting this slider allows you to choose how relevant your data should be before giving it to the bot.\n\nIf your bot is struggling to answer questions about your data sources, try a lower document relevance value to see if that helps. Likewise, if you think your bot is answering questions it shouldn’t be, try increasing the document relevance.\n\n## Customising the bot prompt\n\nThe prompt is the instructions given to your bot that tells it how it should act and how it should answer questions. By changing the prompt, you can get your bot to behave very differently. This is the default prompt we provide to your bot:\n\n> Answer my questions only using data from the included context below in markdown format. Include any relevant media or code snippets. If the answer is not in the text, say I do not know.\n> \n\nYou can change this using the “Use our default prompt” toggle in the bot general settings\n\n![prompt.png](https://res.cloudinary.com/djyjvrw5u/image/upload/v1707322616/prompt_f8991b04c9.png)\n\n### Changing your bots tone\n\nYou can use the prompt to easily change your bots tone. For example, you could say:\n\n> Answer my questions using the data included in the style of a pirate of the high seas.\n\nOr\n\n> Answer my questions using the data included in the style Donald Trump.\n\n### An example prompt template\n\nThe prompt is a very powerful tool in tailoring how you want your bot to behave and respond. Firstly, we want to stress that there are many great resources online to help with getting the most out of your prompt. However, to get you started and to pass on a structure which has worked quite well, we've included an example template below. Some general guidelines as to what each part should contain and examples for each. Please do tailor these to your particular use case and remember this prompt is provided with each query, so the longer it is, the more tokens will be used per query.\n\n**Start by describing how you want the bot to behave, its personality, its role & what it is intended for.**\n\nYou are a helpful assistant, please respond in a friendly yet professional manner.\n\n{% callout title=\"System message\" %}\nYou can provide a shorter version of this in the system message too. From our example: 'You are a helpful assistant' is probably a good example of what you could choose to include for the system message.\n{% /callout %}\n\n**Next, make it clear what kind of information it will be receiving. Specifying Markdown as the input allows us to include images, links, and other structures from your data sources to help your bot respond to users.**\n\nYou will be given information to help answer this question in markdown format.\n\n**Then we explain Chat GPT should reference information identified within your data sources to help its responses. The **\u003Ccontext>** marker is very important, this is where relevant parts of your data source will be included automatically. If you don't include this marker, the context from your data sources will be added after your prompt.**\n\nPlease answer questions using the following context only:\n\u003Ccontext>\n\n**Finally, we list important considerations or restrictions. Asking the bot to respond using markdown allows it to include images/links from your data sources.**\n\nRules:\n- If the answer isn't clear from the provided context above, please say \"I don't know\"\n- Please don't mention phrases like \"the text mentions\", \"according to the context\"\n- Your answers should be short and concise\n- Please respond using markdown format\n\n#### Putting it all together\n\n```\nYou are a helpful assistant, please respond in a friendly yet professional manner.\nYou will be given information to help answer this question in markdown format.\nPlease answer questions using the following context only: \u003Ccontext>\nRules:\n- If the answer isn't clear from the provided context above, please say \"I don't know\"\n- Please don't mention phrases like \"the text mentions\", \"according to the context\"\n- Your answers should be short and concise\n- Please respond using markdown format\n```\n\n## Changing the max context size\n\nEach model has a different context size, this is the maximum number of tokens we can send to ChatGPT to respond to a user's request. Reducing this will reduce the maximum number of tokens each request can use, but it also limits the amount of information we can include from your data sources.\n\n## Changing max context documents\n\nWhen you upload data to Chat Thing, the information is broken down into chunks, we refer to these as **documents**. You can use this slider to control the maximum number of documents we provide to Chat GPT along with each request. Reducing this will limit the number of documents provided, which can help reduce the number of tokens used for each request.\n\n## Changing max message history\n\nThis slider allows you to control the number of previous messages (including user queries and bot responses) provided to Chat GPT. Decreasing the value will limit the amount of previous messages provided, which can help reduce the number of tokens used for each request.","2023-06-16T20:30:43.603Z","2024-07-19T08:56:18.333Z","2023-06-16T20:30:44.974Z",{"data":231},[],[233,239,244,252,257,261,265,270,277,281,286,291,296,300,306,310,316,320,327,331],{"id":234,"name":235,"type":236,"contextSize":237,"initialContextSize":222,"inputTokenMultiplier":6,"outputTokenMultiplier":20,"supportsPowerUps":238,"supportsVision":238,"default":238,"description":222},"1cada48d-5e4a-40c3-ade2-48c7e46ee0ff","Cohere - Command R","cohereCommandR",128000,false,{"id":240,"name":241,"type":242,"contextSize":237,"initialContextSize":222,"inputTokenMultiplier":17,"outputTokenMultiplier":243,"supportsPowerUps":238,"supportsVision":238,"default":238,"description":222},"335c0c8c-8155-46d8-a4ac-8379d5e45af4","Cohere - Command R+","cohereCommandRPlus",30,{"id":245,"name":246,"type":247,"contextSize":248,"initialContextSize":222,"inputTokenMultiplier":249,"outputTokenMultiplier":250,"supportsPowerUps":251,"supportsVision":251,"default":238,"description":222},"fab1297f-9fbd-47d4-825a-97385e0ffbb6","Claude 3 - Haiku","claude3Haiku",200000,0.5,2.5,true,{"id":253,"name":254,"type":255,"contextSize":248,"initialContextSize":222,"inputTokenMultiplier":243,"outputTokenMultiplier":256,"supportsPowerUps":251,"supportsVision":251,"default":238,"description":222},"d94e93df-064e-4835-b0de-f926f4a7e80a","Claude 3 - Opus","claude3Opus",150,{"id":258,"name":259,"type":260,"contextSize":248,"initialContextSize":222,"inputTokenMultiplier":17,"outputTokenMultiplier":243,"supportsPowerUps":251,"supportsVision":251,"default":238,"description":222},"47c4fbee-071b-4b25-9511-8e5a4ea587ad","Claude 3 - Sonnet","claude3Sonnet",{"id":262,"name":263,"type":264,"contextSize":237,"initialContextSize":222,"inputTokenMultiplier":94,"outputTokenMultiplier":243,"supportsPowerUps":251,"supportsVision":251,"default":238,"description":222},"8d72720e-b0e1-4f46-94df-50f4d35b0bfc","GPT-4o","gpt4o",{"id":266,"name":267,"type":268,"contextSize":269,"initialContextSize":222,"inputTokenMultiplier":6,"outputTokenMultiplier":20,"supportsPowerUps":251,"supportsVision":238,"default":238,"description":222},"05bb60e1-665a-4802-9b18-a0ae2be12c12","GPT-3.5 Turbo","gpt3",16384,{"id":271,"name":272,"type":273,"contextSize":274,"initialContextSize":222,"inputTokenMultiplier":275,"outputTokenMultiplier":276,"supportsPowerUps":251,"supportsVision":238,"default":238,"description":222},"712bf42d-75e7-46e2-b41c-9fbdac54bff8","GPT-4","gpt4",8192,60,120,{"id":278,"name":279,"type":280,"contextSize":237,"initialContextSize":222,"inputTokenMultiplier":191,"outputTokenMultiplier":275,"supportsPowerUps":251,"supportsVision":251,"default":238,"description":222},"ac1febbf-5bc4-40fa-b323-02fc76659329","GPT-4 Turbo 128k","gpt4128k",{"id":282,"name":283,"type":284,"contextSize":285,"initialContextSize":222,"inputTokenMultiplier":249,"outputTokenMultiplier":249,"supportsPowerUps":251,"supportsVision":238,"default":238,"description":222},"cef7ac7f-cba5-4b90-b15a-3b6ec4a1f13e","Mistral - Open Mistral 7b","openMistral7b",32000,{"id":287,"name":288,"type":289,"contextSize":285,"initialContextSize":222,"inputTokenMultiplier":290,"outputTokenMultiplier":290,"supportsPowerUps":238,"supportsVision":238,"default":238,"description":222},"7d78d44b-91d1-4033-aae8-bc877dfe7cc0","Mistral - Open Mixtral 8x7b","openMixtral8x7b",1.4,{"id":292,"name":293,"type":294,"contextSize":295,"initialContextSize":222,"inputTokenMultiplier":22,"outputTokenMultiplier":22,"supportsPowerUps":251,"supportsVision":238,"default":238,"description":222},"e952542a-f774-4a3a-87d2-fd699e9156ac","Mistral - Open Mixtral 8x22b","openMixtral8x22b",64000,{"id":297,"name":298,"type":299,"contextSize":285,"initialContextSize":222,"inputTokenMultiplier":13,"outputTokenMultiplier":17,"supportsPowerUps":251,"supportsVision":238,"default":238,"description":222},"3487fd65-7fa1-4954-869a-86691367da71","Mistral - Mistral Small","mistralSmall",{"id":301,"name":302,"type":303,"contextSize":285,"initialContextSize":222,"inputTokenMultiplier":304,"outputTokenMultiplier":305,"supportsPowerUps":251,"supportsVision":238,"default":238,"description":222},"8a1438bf-53ef-458f-bc3a-0a9ddc561b76","Mistral - Mistral Medium","mistralMedium",5.4,16.2,{"id":307,"name":308,"type":309,"contextSize":285,"initialContextSize":222,"inputTokenMultiplier":75,"outputTokenMultiplier":34,"supportsPowerUps":251,"supportsVision":238,"default":238,"description":222},"4d8e3a35-9dc0-4c16-8a6f-9e88f7056375","Mistral - Mistral Large","mistralLarge",{"id":311,"name":312,"type":313,"contextSize":237,"initialContextSize":269,"inputTokenMultiplier":314,"outputTokenMultiplier":315,"supportsPowerUps":251,"supportsVision":251,"default":251,"description":222},"9852023e-fe6a-4b91-818d-b64e2cbaa3ae","GPT-4o Mini","gpt4omini",0.3,1.2,{"id":317,"name":318,"type":319,"contextSize":248,"initialContextSize":222,"inputTokenMultiplier":17,"outputTokenMultiplier":243,"supportsPowerUps":251,"supportsVision":238,"default":238,"description":222},"997cae51-b852-4260-81c2-1a7833ddcaaa","Claude 3.5 - Sonnet","claude35Sonnet",{"id":321,"name":322,"type":323,"contextSize":324,"initialContextSize":222,"inputTokenMultiplier":325,"outputTokenMultiplier":326,"supportsPowerUps":251,"supportsVision":251,"default":238,"description":222},"418f12dd-a32d-4caf-8d73-2b6fabe9fdb7","Google - Gemini 1.5 Flash","gemini15Flash",1000000,0.7,2.1,{"id":328,"name":329,"type":330,"contextSize":324,"initialContextSize":222,"inputTokenMultiplier":37,"outputTokenMultiplier":195,"supportsPowerUps":251,"supportsVision":251,"default":238,"description":222},"1b18d61f-9824-44fc-8252-a99058b68528","Google - Gemini 1.5 Pro","gemini15Pro",{"id":332,"name":333,"type":334,"contextSize":248,"initialContextSize":222,"inputTokenMultiplier":13,"outputTokenMultiplier":94,"supportsPowerUps":251,"supportsVision":251,"default":238,"description":222},"030d2041-c4f1-4d08-b85e-12c5d5ecfb91","Claude 3.5 - Haiku","claude35Haiku",["Reactive",336],{"$ssupabase_session":222,"$ssupabase_user":222,"$snuxt-delay-hydration-mode":337,"$snuxt-i18n-meta":338,"$scolor-mode":339,"$sauthUser":-1,"$ssite-config":341},"mount",{},{"preference":340,"value":340,"unknown":251,"forced":238},"dark",{"currentLocale":208,"defaultLocale":208,"env":342,"name":343,"url":344},"production","chatthing","https://chatthing.ai",["Set"],["ShallowReactive",347],{"docs.navigation":-1,"docs.available-models":-1,"Gv5U0kKU2Q":-1},"/docs/available-models",["Reactive",350],{"docs":351,"notification":367},{"headings":352},["Ref",353],["Reactive",354],[355,359,363],{"id":356,"title":357,"children":358},"available-models-and-capabilities","Available models and capabilities",[],{"id":360,"title":361,"children":362},"model-pricing","Model pricing",[],{"id":364,"title":365,"children":366},"changing-the-model","Changing the model",[],{"notifications":368,"position":371},["Ref",369],["Reactive",370],[],["Ref",372],"bottom"]