The main AI models in September 2026 are OpenAI’s GPT-6 family (Astra, Sol and Luna), Anthropic’s Claude Opus 5.5, Google’s Gemini 3.1 Pro and 3.8 Flash, xAI’s Grok 4.7, Meta’s Muse Spark and open-weight options such as DeepSeek V4.1 and Qwen. For images, video and voice, Nano Banana, Midjourney, Veo, Sora, Runway, Kling and ElevenLabs lead. This guide explains what each is for, where to use it and what it costs.
You do not need the “best” model. You need one that fits the job, your budget and the tools you already use. The honest truth is that for everyday writing, summarising and planning, every major assistant is good enough – the prompt matters more than the model. The differences show up at the edges: very long documents, complex code, images, video and price at volume.
Not sure where to start?
Tell us what you want to do and what you can spend. Get a recommendation in ten seconds.
Checked against providers' official pages on September 25, 2026. Models change fast - follow the links for current details.
Chat, writing and reasoning models
GPT-6 Astra
OpenAI · Sep 2026OpenAI's flagship: complex professional work, coding, research and computer-use tasks.
- Where to use it
- ChatGPT paid plans; API
- Price
- $10 / $50 per 1M tokens (API, standard)
GPT-6 Sol
OpenAI · Sep 2026Cost-efficient model for everyday work with much of Astra's capability.
- Where to use it
- ChatGPT paid plans; API
- Price
- $2 / $10 per 1M tokens (API)
GPT-6 Luna
OpenAI · Sep 2026Fast, very low-cost model for high-volume and simple tasks.
- Where to use it
- ChatGPT (incl. free/Go on desktop); API
- Price
- $0.10 / $0.50 per 1M tokens (API)
Claude Opus 5.5
Anthropic · Sep 2026Anthropic's most capable generally available model: agentic coding, long knowledge work, computer use.
- Where to use it
- Claude paid plans; API; AWS, Google Cloud, Azure
- Price
- $4 / $20 per 1M tokens (API)
Claude Sonnet & Haiku
Anthropic · OngoingFaster, lower-cost Claude tiers for everyday writing, chat and high-volume work. Anthropic has announced 5.5 versions as coming next.
- Where to use it
- Claude (free and paid); API
- Price
- See Anthropic pricing
Gemini 3.1 Pro
Google · 2026Google's model for complex tasks and creative projects, with strong multimodal (image, video, audio) understanding.
- Where to use it
- Gemini app; Google AI Studio / Vertex AI
- Price
- See Google pricing
Gemini 3.8 Flash
Google · Sep 2026Fast model for agentic tasks at scale, long documents and multimodal input.
- Where to use it
- Gemini app; API
- Price
- See Google pricing (introductory rates announced)
Gemini 3.1 Deep Think
Google · 2026Extended reasoning mode for maths, science and engineering problems.
- Where to use it
- Gemini paid plans
- Price
- Subscription
Grok 4.7
xAI · Sep 2026xAI's latest model, positioned for chat and coding, with access to real-time posts on X.
- Where to use it
- Grok app and X; API
- Price
- $2 / $6 per 1M tokens reported at launch
Muse Spark
Meta · 2026The model family behind Meta AI across WhatsApp, Instagram, Facebook and the Meta AI app.
- Where to use it
- Meta AI (free in Meta apps)
- Price
- Free in apps
DeepSeek V4.1
DeepSeek · Sep 2026Low-cost model with open weights (V4.1 Flash under an MIT licence) - popular for self-hosting and budget API use.
- Where to use it
- DeepSeek app; API; open weights
- Price
- From about $0.15 per 1M input tokens (Flash)
Qwen 3.x family
Alibaba · 2026Large family of models, many with open weights, strong in multilingual and coding tasks.
- Where to use it
- Qwen Chat; API; open weights
- Price
- Varies by model
Mistral models
Mistral AI · 2026European provider with open-weight and commercial models; Le Chat assistant.
- Where to use it
- Le Chat; API; open weights
- Price
- Varies by model
Image models
Nano Banana (Gemini Image)
Google · 2025-26Image generation and precise photo editing from text instructions - behind many viral photo trends.
- Where to use it
- Gemini app; API
- Price
- Free tier in app; API pricing
ChatGPT image generation
OpenAI · OngoingConversational image creation and editing inside ChatGPT, good at following detailed instructions and text in images.
- Where to use it
- ChatGPT
- Price
- Included with plan limits
Midjourney (V8 series)
Midjourney · 2026Highly aesthetic, stylised images; strong for concept art, mood and illustration.
- Where to use it
- Web app (paid)
- Price
- Subscription
Adobe Firefly
Adobe · OngoingImage, vector, video and sound-effect generation integrated with Photoshop and Adobe Express; designed for commercial use.
- Where to use it
- Web; Adobe apps
- Price
- Free tier; Adobe plans
Video models
Veo
Google · OngoingCinematic text-to-video and image-to-video with generated audio.
- Where to use it
- Gemini app; Flow; API
- Price
- Paid plans / API
Sora
OpenAI · OngoingText-to-video and remixing short clips with sound.
- Where to use it
- Sora app; ChatGPT plans
- Price
- Plan-based
Runway
Runway · OngoingVideo generation and editing tools for creators and studios.
- Where to use it
- Web app
- Price
- Free trial; subscription
Kling
Kuaishou · OngoingPopular text- and image-to-video model for realistic motion.
- Where to use it
- Web app
- Price
- Free credits; subscription
Voice and music
ElevenLabs
ElevenLabs · OngoingNatural text-to-speech voices, dubbing and sound effects - common for faceless YouTube narration.
- Where to use it
- Web; API
- Price
- Free tier; subscription
Suno
Suno · OngoingGenerates full songs with vocals from a text prompt.
- Where to use it
- Web; app
- Price
- Free tier; subscription
At a glance
| Model | Maker | Type | Good for | API price (in / out per 1M tokens) |
|---|---|---|---|---|
| GPT-6 Astra | OpenAI | Chat & writing | writing, coding, analysis | $10 / $50 per 1M tokens (API, standard) |
| GPT-6 Sol | OpenAI | Chat & writing | writing, coding, analysis | $2 / $10 per 1M tokens (API) |
| GPT-6 Luna | OpenAI | Chat & writing | writing, bulk, chat | $0.10 / $0.50 per 1M tokens (API) |
| Claude Opus 5.5 | Anthropic | Chat & writing | writing, coding, analysis | $4 / $20 per 1M tokens (API) |
| Claude Sonnet & Haiku | Anthropic | Chat & writing | writing, chat, bulk | See Anthropic pricing |
| Gemini 3.1 Pro | Chat & writing | writing, analysis, research | See Google pricing | |
| Gemini 3.8 Flash | Chat & writing | bulk, analysis, multimodal | See Google pricing (introductory rates announced) | |
| Gemini 3.1 Deep Think | Deep reasoning | research, analysis, maths | Subscription | |
| Grok 4.7 | xAI | Chat & writing | chat, coding, news | $2 / $6 per 1M tokens reported at launch |
| Muse Spark | Meta | Chat & writing | chat, writing, social | Free in apps |
| DeepSeek V4.1 | DeepSeek | Open-weight & budget | coding, bulk, open-weights | From about $0.15 per 1M input tokens (Flash) |
| Qwen 3.x family | Alibaba | Open-weight & budget | coding, multilingual, open-weights | Varies by model |
| Mistral models | Mistral AI | Open-weight & budget | writing, open-weights, multilingual | Varies by model |
| Nano Banana (Gemini Image) | Image | image, photo-editing, viral | Free tier in app; API pricing | |
| ChatGPT image generation | OpenAI | Image | image, photo-editing, viral | Included with plan limits |
| Midjourney (V8 series) | Midjourney | Image | image, art | Subscription |
| Adobe Firefly | Adobe | Image | image, design, commercial | Free tier; Adobe plans |
| Veo | Video | video | Paid plans / API | |
| Sora | OpenAI | Video | video, social | Plan-based |
| Runway | Runway | Video | video, editing | Free trial; subscription |
| Kling | Kuaishou | Video | video, social | Free credits; subscription |
| ElevenLabs | ElevenLabs | Voice & music | voice, audio, youtube | Free tier; subscription |
| Suno | Suno | Voice & music | music, audio | Free tier; subscription |
How to choose, in plain English
- Everyday writing, email and planning: any of ChatGPT, Claude or Gemini on a free plan. Choose the one inside the apps you already use.
- Long documents, careful writing and coding: Claude and the top GPT-6 and Gemini models are the usual picks. Try the same prompt in two and keep the one you prefer.
- Research with sources: an assistant with web search (Perplexity, ChatGPT, Gemini, Claude with search) – then open the sources and check them.
- Images and photo editing: Nano Banana in Gemini or image generation in ChatGPT for edits of your own photos; Midjourney for stylised art; Firefly for commercial design work.
- Video: Veo, Sora, Runway or Kling for short clips. Expect to generate several versions and edit them together.
- Voice and music: ElevenLabs for narration, Suno for songs. Check commercial-use terms before publishing.
- High volume on a budget (API): small, cheap models such as GPT-6 Luna, Gemini Flash-Lite or DeepSeek V4.1 Flash. Estimate costs with the Token & Cost Calculator.
Model names change every few months
The model behind a product can change without the product name changing. A prompt that works well today may behave differently after an update. Keep your best prompts saved, and re-check important outputs after a model change.
Questions people ask
What is the best AI model right now?
There is no single best model. OpenAI, Anthropic and Google each release a new top model every few months, and each leads on different tasks. For most people the right answer is the assistant they find easiest to use, with a good prompt.
Is ChatGPT, Claude or Gemini better?
All three are strong general assistants. Many people use ChatGPT for all-round tasks and images, Claude for long documents, writing and code, and Gemini inside Google apps and for photo editing. Try the same prompt in two and compare.
Which AI is free?
ChatGPT, Claude, Gemini, Meta AI, Copilot, Grok and DeepSeek all have free plans with usage limits. Paid plans add stronger models, higher limits and extra features.
What does open weights mean?
The model’s trained weights are published so anyone can run it on their own hardware or cloud, within the licence. DeepSeek, Qwen and Mistral offer open-weight models.
How much does the API cost?
API prices are per million tokens (roughly 750,000 words) and range from about ten cents to tens of dollars depending on the model. Use the Token & Cost Calculator for an estimate.