What the compute is for
33 current models from 10 providers: when each came out, what it can do, what it costs to run, and what a subscription costs where you live. Every figure links to where it came from.
OpenAI
“Our most capable model for the most demanding work.”
$10 in · $50 out per million tokens
What people build with it T2C summary
OpenAI
“Near-Astra performance for complex work at a lower cost.”
$2 in · $10 out per million tokens
What people build with it T2C summary
OpenAI
“Our most efficient model for focused, high-volume tasks.”
$0.1 in · $0.5 out per million tokens
What people build with it T2C summary
OpenAI
“Our most capable model for image generation and editing.”
$8 in · $30 out per million tokens
What people build with it T2C summary
OpenAI
“Our premier model for natural, expressive voice conversations with smooth interruption handling.”
$0.05 a minute
What people build with it T2C summary
Anthropic
“For demanding reasoning and long-horizon agentic work”
$10 in · $50 out per million tokens
What people build with it T2C summary
Anthropic
“For long-running agentic coding and knowledge work”
$4 in · $20 out per million tokens
What people build with it T2C summary
Anthropic
“The best combination of speed and intelligence”
$2 in · $10 out per million tokens
What people build with it T2C summary
Anthropic
“For high-volume, latency-sensitive tasks such as classification, extraction, and routing”
From $0.1 in · $0.5 out per million tokens
What people build with it T2C summary
“Our flagship reasoning model with a unified 1M context window and long decodes up to 1M tokens.”
$2 in · $10 out per million tokens
What people build with it T2C summary
“Our most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents…”
$0.75 in · $3.75 out per million tokens
What people build with it T2C summary
“High-efficiency image generation and editing model with improved visual quality, typography, and multi-turn consistency.”
$0.0336 an image
What people build with it T2C summary
“Fast video generation, editing, keyframe interpolation, and extension with native audio.”
$0.1 a second
What people build with it T2C summary
“Our flagship music generation model, optimized for full-length songs with complex structural coherence.”
$0.08 a song
What people build with it T2C summary
Meta
“Frontier coding, long-running agents, and advanced computer use.”
$4 in · $20 out per million tokens
What people build with it T2C summary
Meta
“Built for long-horizon coding and agentic work.”
$1.25 in · $4.25 out per million tokens
What people build with it T2C summary
Meta
“The most capable US open-weight model to date.”
$1.25 in · $4.25 out per million tokens
What people build with it T2C summary
Meta
“Detect, segment, and track objects from a text prompt, across image and video.”
API price not published
What people build with it T2C summary
Microsoft
“Competitive reasoning and top SWE-Bench Pro results at a mid-weight price.”
$2 in · $8 out per million tokens
What people build with it T2C summary
Microsoft
“A lightweight, agentic model built into GitHub Copilot and VS Code.”
API price not published
What people build with it T2C summary
Microsoft
“Holds up across long generations.”
API price not published
What people build with it T2C summary
Microsoft
“Design-ready images from any text or photo prompt.”
API price not published
What people build with it T2C summary
xAI
“For everything else, including code, use Grok 4.7. It is the most capable model we've built.”
$2 in · $6 out per million tokens
What people build with it T2C summary
xAI
xAI recommends it for video.
$0.08 a second
What people build with it T2C summary
xAI
Speech-to-speech model in the Grok Voice API.
$0.08 a minute
What people build with it T2C summary
DeepSeek
“Stronger benchmark performance, faster task completion, and shorter outputs.”
$0.15 in · $0.6 out per million tokens
What people build with it T2C summary
DeepSeek
Listed as a current API model; DeepSeek gives no description.
$0.66 in · $1.98 out per million tokens
What people build with it T2C summary
Mistral AI
“A state-of-the-art, open-weight, general-purpose multimodal model.”
$1.36 in · $4.18 out per million tokens
What people build with it T2C summary
Mistral AI
“State-of-the-art TTS with zero-shot and professional voice cloning, 9 languages, and streaming.”
API price not published
What people build with it T2C summary
Mistral AI
“Our latest OCR service with paragraph-level bounding boxes, structural block labels, and block-level confidence scores.”
API price not published
What people build with it T2C summary
Alibaba Qwen
Listed first among Model Studio's text and vision models.
$2 in · $6 out per million tokens
What people build with it T2C summary
Alibaba Qwen
Model Studio's real-time audio and video conversation model.
$0.15 in · $0.47 out per million tokens
What people build with it T2C summary
NVIDIA
“Combines text and visual understanding to help AI agents extract information from webpages, PDFs, tables, and charts.”
$0.15 in · $0.5 out per million tokens
What people build with it T2C summary
Marketplace estimate means the provider publishes no price or date, so the figure is from OpenRouter, the largest marketplace reselling model APIs. Where both exist they usually match, but OpenRouter can charge more. Its date is when it listed the model. “What people build with it” is T2C's own summary of the provider's description.
The App Store price is what each provider charges in that country's App Store, read 25 Oct 2026. It includes local tax and can be higher than buying on the web, because Apple takes a share. The web price is from the provider's own pricing page, in the currency it showed.
| Plan | App Store | Web | Includes |
|---|---|---|---|
| Free | — | $0 a month | GPT-6 Luna |
| Go | $8.00 ChatGPT Go | £7 a month | GPT-6 Luna, with more messages |
| Plus | $19.99 ChatGPT Plus | $20 · £20 a month | GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna |
| Pro | $100.00 ChatGPT Pro 100 | from $100 · from £89 a month | “Pro reasoning powered by GPT-6 Astra”; three usage tiers |
| Plan | App Store | Web | Includes |
|---|---|---|---|
| Free | — | $0 a month | Sonnet, Haiku |
| Pro | $20.00 Claude Pro - Monthly | $20 a month | Opus, Sonnet, Haiku; Fable through usage credits; Claude Code |
| Max 5x | $124.99 Claude Max 5x - Monthly | from $100 a month | Everything in Pro; Fable at 50% of weekly limits |
| Max 20x | $249.99 Claude Max 20x - Monthly | Not published | As Max 5x, with 20x Pro's usage |
| Plan | App Store | Web | Includes |
|---|---|---|---|
| Free | — | £0 a month | Flash-Lite, Deep Research, Gemini Live |
| AI Plus | $4.99 Google AI Plus (400 GB) | £4.49 a month | Flash; 400 GB storage |
| AI Pro | $19.99 Google AI Pro (5 TB) | £18.99 a month | Gemini 4 Argon, Deep Research, Omni Flash; 5 TB storage |
| AI Ultra | $199.99 Google AI Ultra (30 TB) | from £79.99 a month | Highest Gemini 4 Argon access; Veo 3.1; from 20 TB storage |
| Plan | App Store | Web | Includes |
|---|---|---|---|
| SuperGrok Lite | $10.00 SuperGrok Lite | Not published | Not itemised by xAI |
| SuperGrok | $30.00 SuperGrok | Not published | “Higher limits and multi-agent reasoning” |
| SuperGrok Plus | $100.00 SuperGrok Plus | Not published | Not itemised by xAI |
| SuperGrok Heavy | $300.00 SuperGrok Heavy | Not published | Not itemised by xAI |
| Plan | App Store | Web | Includes |
|---|---|---|---|
| Free | — | $0 a month | Limited messages and coding sessions |
| Pro | $14.99 Vibe Pro | $14.99 a month | More messages, all-day coding, $15 a month in API credits |
| Team | — | $24.99 a seat a month | Higher limits, 30 GB storage per user |
| Plan | App Store | Web | Includes |
|---|---|---|---|
| Personal | $9.99 Microsoft 365 Personal | $9.99 a month | Copilot features; Microsoft names no model |
| Family | $12.99 Microsoft 365 Family | $12.99 a month | Copilot for the subscription owner only |
| Premium | $19.99 Microsoft 365 Premium Monthly | $19.99 a month | Copilot features; Microsoft names no model |
Provider documentation, changelogs, launch posts and pricing pages, read 25 Oct 2026; the lineup itself as listed on 25 Oct 2026. Prices change often, so check the linked page before you buy. No provider says which model runs in which data centre, so T2C never links a model to a site.