Choose Your AI Model in Jobbit: Claude, GPT, Gemini, DeepSeek, Kimi and More in One Agent

Every serious AI user has a private ranking. Claude for writing and code, GPT for breadth, Gemini for long documents, DeepSeek when the bill matters, Kimi for very long files. Until now, acting on that ranking meant juggling subscriptions and copying text between tabs. Jobbit now lets you choose the model behind your agent, per chat, from a catalogue that includes Jobbit's own tiers and named third-party models, and on paid plans it also lets you choose which models generate your images, video, music and speech.

This guide explains how model selection in Jobbit works: the Jobbit 1.13 Lite, Pro and Ultra tiers, the named models from Anthropic, OpenAI, Google, DeepSeek, Moonshot, xAI, Alibaba, Meta and Z.ai, what the free plan includes and what needs a subscription, how the generation model settings work, and a practical guide to which model to pick for which job. For the wider landscape of assistants, see best ChatGPT alternatives; this post is about using them all from one place.

Why a multi-model agent beats another subscription

Jobbit is a multipurpose AI agent: one chat that researches, writes, builds and deploys software, generates media and runs automations, with a team of subagents behind it (see AI subagents explained). The model is the reasoning engine inside that machine. Being able to swap the engine matters for three reasons:

• Different models are good at different things. Coding, prose, analysis, multilingual work and long-document reading each have a leader, and the leader changes every few months.

• Cost. A budget model can run a routine automation at a fraction of the price of a frontier model. Choosing per chat puts that saving in your hands.

• Continuity. Your projects, files, automations and hosted apps stay in Jobbit whatever model you pick. Switching models does not mean switching tools.

Multi-model access is what products like OpenRouter, Poe and TypingMind offer for chat. Jobbit offers it inside an agent that finishes the work.

The Jobbit 1.13 tiers

Every chat starts on one of Jobbit's own tiers. They are tuned for agent work, tool use, planning and long runs, and the tier decides how much capability you bring to a task:

You pick the tier from the model button in the composer, and the choice applies to that chat. If a plan lapses while a paid tier is selected, the chat falls back to Lite rather than failing.

Named third-party models

Next to the tiers sits a list of named models, each shown with its provider's mark, its context window, its maximum output and its price per million tokens in and out, so you can see what you are choosing rather than guessing. The catalogue currently draws on models from:

• Anthropic, the Claude family, widely preferred for writing quality and agentic coding.

• OpenAI, the GPT family, the broadest all-rounder.

• Google, Gemini, strong on very long context and multimodal input.

• DeepSeek, near-frontier reasoning at a fraction of the cost.

• Moonshot AI, Kimi, known for long documents and coding.

• xAI, Grok, with a distinctive style and real-time strengths.

• Alibaba, Qwen, excellent multilingual quality and a range of sizes.

• Meta, Llama, the open-weight workhorse.

• Z.ai, the GLM family, a capable open-weight line.

Two of the named models are open to every account, including Free; the rest are part of a paid plan. The catalogue is served live, so new models appear as they are added and the list in your composer is always the current one. Model names move fast, which is why the picker shows each model's actual specifications rather than relying on reputation.

Choosing generation models: images, video, music and speech

Text is half the story. Jobbit generates images, video, music and speech as part of jobs, product photos for a shop, a launch video, a jingle, a voiceover, and on paid plans you can now choose the model behind each of those roles in Settings under Models.

• Images. Options from Black Forest Labs (the Flux family), Recraft, Krea, Google, OpenAI, ByteDance and others, each with its price in credits per image.

• Video. Models from Kling, Runway, MiniMax, ByteDance and Google, priced per few seconds of output.

• Music and speech. Generation and voice models with per-unit pricing.

Each role shows what runs by default, what you have chosen and what will actually run, and every model lists its price in credits so you can trade quality against cost deliberately. Free accounts keep running on the defaults, which are chosen to be good rather than cheap; choosing is part of a paid plan. Our guide to AI product photo generators shows what the image side is used for.

What each plan includes

Plan prices and hosting allowances are in Jobbit pricing explained. Credits are the usage currency; a frontier model on a long run costs more credits than Lite on a short one, and the picker's per-million prices tell you roughly how much more.

Which model for which job

A working rule of thumb, to be revised as the catalogue changes:

• Everyday tasks and automations: Jobbit 1.13 Lite or a budget named model such as DeepSeek or Qwen. Fast, cheap, good enough for routine work.

• Writing that has to read well: a Claude model, or Jobbit 1.13 Pro, which is tuned for documents and research.

• Building and debugging software: Jobbit 1.13 Ultra or a top Claude or GPT model; the tiers are tuned for the tool use that app building needs. See best vibe coding tools.

• Very long documents and datasets: Gemini or Kimi for context length, or Jobbit 1.13 Pro with a project holding the files, described in Jobbit Projects.

• Multilingual work: Qwen or a frontier model; check output on your language before committing a batch.

• Hard analysis and long autonomous runs: Jobbit 1.13 Ultra. Pair it with Goal mode, explained in steer your AI agent mid-run.

When in doubt, start on a tier, look at the result, and switch models in the same chat. The agent keeps the context.

How this compares with using the vendors directly

For most small businesses and founders the question is not which chatbot but which engine to put in the agent that does the work. Now you can choose.

Frequently asked questions

Can I use Claude, GPT and Gemini inside Jobbit?

Yes. Paid plans include named models from Anthropic, OpenAI, Google and others alongside Jobbit's own 1.13 tiers, selectable per chat from the model button. Two named models are open to free accounts as well. The catalogue is live, so it reflects the models currently available.

What are Jobbit 1.13 Lite, Pro and Ultra?

They are Jobbit's own model tiers tuned for agent work. Lite is free and suits quick, simple tasks; Pro delivers high quality on complex tasks such as research and app builds; Ultra is the most capable tier for the most demanding jobs. Pro and Ultra are part of paid plans.

Do I need a paid plan to choose models?

Choosing tiers and named models beyond the free ones, and choosing generation models for images, video, music and speech, is part of the Plus, Pro and Max plans. Free accounts run on Jobbit 1.13 Lite, two open named models and the default generation models.

How much do different models cost in Jobbit?

The picker shows each named model's price per million tokens in and out, and each generation model's price in credits per image or per few seconds of video, so you can compare before choosing. A frontier model on a long task uses more credits than Lite on a short one. Plan credit allowances are on the pricing page.

Can I change model in the middle of a chat?

Yes. Pick a different model from the composer and the same chat, project files and context continue with the new engine. If a paid model becomes unavailable because a plan lapses, the chat falls back to Lite.

Pick the engine, keep the agent. Start free on Jobbit and open the model menu in your first chat.