Imagera AI - AI content creation platform for generating images, cloning voices, and enhancing videos. Privacy Policy | Terms

Illustration: a developer shipping late at nightIllustration

Imagera Code · coding agent for your terminal

Stop rationing your coding agent.

No 3-hour lockouts.

Describe the change. It reads your project, edits the files, runs your tests and shows you the diff, on any of 280 coding models, from one plan.

npm i -g imagera-code

From $20/month · cancel any time · past a limit, keep going on credits.

  • 280coding models, updated live
  • 0short lockouts: one weekly limit only
  • $0.66a day on Pro
  • 0lines of your code kept by Imagera

How to use

Watch it fix a failing build.

A real 32-second run on a demo project, sped up while the model thinks. Jump to any step.

Real recording, sped up while the model thinks. Two failing tests, fixed on the default model.
Illustration: shipping before sunriseIllustration

Ship it before sunrise.

Hand it the bug at midnight. Read the diff with your coffee.

Illustration: friends building together at a hackathonIllustration

Every model

Every frontier model. One plan.

Switch mid-chat with /models.

Claude Opus, Claude Fable, GPT-6, Gemini, Grok and DeepSeek and the rest: 280 coding models from 41 labs, all on one plan. New ones appear on their own, and you never pay a second subscription to try one.

  • Claude Opus7 models
  • Claude Fable3 models
  • GPT-610 models
  • Gemini17 models
  • Grok6 models
  • DeepSeek13 models
  • GLM 5.3 Flash
  • Qwen4 Max
  • ERNIE-Light-DT-1.0
  • Nex-N2.6
  • Claude Fable 5.2
  • LFM3-1.7B
  • Ripple 65B Preview
  • Motif 3
  • Particle-1
  • GPT-6.1 Terra
  • MiniMax M3.1 Flash
  • Glimpse
  • Ternary Bonsai 2 9B
  • Nemotron 3.5 Super VL
  • Pareto 26.9
  • Gemini 4 Argon
  • Step 5 Preview
  • Mistral Large 4
  • Pareto 26.10 Preview
  • GPT-6.1 Sol
  • Perceptron Mk1.5
  • GLM 5.3 Prime
  • Aion 3.5 Mini
  • Solar Mini 4
  • GPT-6 Luna Pro
  • GPT-6 Sol Pro
  • Claude Opus 5.5
  • MiMo-V2.6-Flash
  • Grok 4.7
  • Ternary Bonsai 2 27B
  • Fugu Ultra v2
  • Ling 3.0 Flash VL
  • Nex-N2.5-Pro
  • GPT-6 Astra Pro
  • Muse Spark 1.3 Contributor
  • Gemini 3.8 Flash
  • Granite 4.2 8B
  • Qwen3.8 Flash
  • DeepSeek V4 Flash Vision Exp
  • Qwen3.8 27B
  • Seed 2.1 Turbo
  • DeepSeek V4 Pro 0813
  • Nemotron 3.5 Lightning
  • Solar Pro 4
  • Ling 3.0 Tiny
  • DeepSeek V4 Flash 0731
  • Qwen3.7 Flash
  • Ling 3.0 Flash
  • Gemini 3.5 Flash Lite
  • Inkling
  • Muse Spark 1.1
  • GPT-5.6 Luna
  • GPT-5.6 Terra
  • GPT-5.6 Sol
  • Aion-3.0-Mini
  • Hy3
  • Fugu Ultra
  • Kimi K2.7 Code
  • Nemotron 3 Ultra
  • MiniMax M3
  • Claude Opus 4.8
  • Grok Build 0.1
  • Gemini 3.1 Flash Lite
  • Grok 4.3
  • Qwen3.5 Plus 2026-04-20
  • Qwen3.6 35B A3B
  • GPT-5.5 Pro
  • DeepSeek V4 Pro 0423
  • Hy3 preview
  • MiMo-V2.5
  • Claude Opus 4.7
  • Gemma 4 26B A4B
  • Qwen3.6 Plus
  • Grok 4.20
  • MiniMax M2.7
  • GPT-5.4 Mini
  • GLM 5 Turbo
  • Seed-2.0-Lite
  • GPT-5.4 Pro
  • Gemini 3.1 Flash Lite Preview
  • Qwen3.5-35B-A3B
  • Qwen3.5-122B-A10B
  • Gemini 3.1 Pro Preview Custom Tools
  • Aion-2.0
  • Claude Sonnet 4.6
  • Qwen3.5 397B A17B
  • GLM 5
  • Qwen3 Coder Next
  • GLM 4.7 Flash
  • GLM 4.7
  • Nemotron 3 Nano 30B A3B
  • GPT-5.2 Pro
  • Devstral 2 2512
  • GLM 4.6V
  • Nova 2 Lite
  • Ministral 3 8B 2512
  • Mistral Large 3 2512
  • Claude Opus 4.5
  • GPT-5.1-Codex
  • Kimi K2 Thinking
  • Voxtral Small 24B 2507
  • Claude Haiku 4.5
  • Qwen3 VL 30B A3B Instruct
  • GLM 4.6
  • DeepSeek V3.2 Exp
  • DeepSeek V3.1 Terminus
  • Qwen3 Next 80B A3B Instruct
  • DeepSeek V3.1
  • GLM 4.5V
  • GPT-5 Mini
  • gpt-oss-120b
  • Claude Opus 4.1
  • Qwen3 Coder 30B A3B Instruct
  • GLM 4.5
  • Qwen3 235B A22B Thinking 2507
  • Gemini 2.5 Flash Lite
  • Kimi K2 0711
  • Gemini 2.5 Flash
  • o3 Pro
  • R1 0528
  • Qwen3 30B A3B
  • Qwen3 32B
  • GPT-4.1
  • GPT-4.1 Nano
  • Llama 4 Scout
  • Mistral Small 3.1 24B
  • Command A
  • Saba
  • DeepSeek V3
  • Llama 3.3 70B Instruct
  • Nova Micro 1.0
  • GPT-4o (2024-11-20)
  • Qwen2.5 72B Instruct
  • Command R+ (08-2024)
  • Llama 3.1 70B Instruct
  • Mistral Nemo
  • GPT-4o-mini (2024-07-18)
  • GPT-4o (2024-05-13)
  • Mistral Large
  • GPT-3.5 Turbo 16k
  • Hy4 flash
  • Gemini 4 Hydrogen
  • Gemini 4 Carbon
  • Inkling 2 Small
  • Muse Ember
  • Mi:dm K 3.0 Pro
  • Hy4
  • GLM 5.3 FlashX Legal
  • GPT-6.1 Terra Pro
  • Kimi K3.1 Pro
  • DeepSeek V4.1 Flash 1017
  • Beam 1.0 501B A23B
  • Muse Spark Open
  • Pareto 26.10
  • LFM2.5-24B-A2B
  • Muse Wonder
  • Claude Haiku 5.5
  • Ling 3.1 Flash
  • GPT-6.1 Sol Pro
  • Claude Sonnet 5.5
  • Ember-1
  • Qwen3.8 Max Prime
  • Aion 3.5
  • Command A+
  • GPT-6 Luna
  • GPT-6 Sol
  • MiMo-V2.6-Pro-UltraSpeed
  • MiMo-V2.6-Pro
  • Qwen3.8 Omni Flash
  • GLM 5.3 FlashX
  • Fugu Max
  • DeepSeek V4.1 Flash
  • GPT-6 Astra
  • Qwen3.8 Max (0902)
  • Muse Spark 1.3
  • Claude Fable 5.1
  • Hy4 preview
  • Muse Spark 1.2 Contributor
  • GLM 5.3
  • Gemini 3.7 Flash
  • Qwen3.8 2.4T A95B
  • Grok 4.6
  • Sakana Namazu
  • Muse Glimmer 30B
  • Muse Spark 1.2
  • Inkling Small
  • Claude Opus 5
  • Gemini 3.6 Flash
  • LongCat 2.0
  • Kimi K3
  • GPT-5.6 Luna Pro
  • GPT-5.6 Terra Pro
  • GPT-5.6 Sol Pro
  • Grok 4.5
  • Aion-3.0
  • Claude Sonnet 5
  • GLM 5.2
  • Claude Fable 5
  • Qwen3.7 Plus
  • Step 3.7 Flash
  • Qwen3.7 Max
  • Gemini 3.5 Flash
  • GPT Chat Latest
  • Mistral Medium 3.5
  • Qwen3.6 Flash
  • Qwen3.6 27B
  • GPT-5.5
  • DeepSeek V4 Flash 0423
  • MiMo-V2.5-Pro
  • Kimi K2.6
  • GLM 5.1
  • Gemma 4 31B
  • GLM 5V Turbo
  • Reka Edge
  • GPT-5.4 Nano
  • Mistral Small 4
  • Nemotron 3 Super
  • Qwen3.5-9B
  • GPT-5.4
  • Seed-2.0-Mini
  • Qwen3.5-27B
  • Qwen3.5-Flash
  • GPT-5.3-Codex
  • Gemini 3.1 Pro Preview
  • Qwen3.5 Plus 2026-02-15
  • MiniMax M2.5
  • Claude Opus 4.6
  • Kimi K2.5
  • GPT-5.2-Codex
  • Gemini 3 Flash Preview
  • GPT-5.2 Chat
  • GPT-5.2
  • Relace Search
  • GPT-5.1-Codex-Max
  • Ministral 3 14B 2512
  • Ministral 3 3B 2512
  • DeepSeek V3.2
  • GPT-5.1
  • GPT-5.1-Codex-Mini
  • Nova Premier 1.0
  • gpt-oss-safeguard-20b
  • Qwen3 VL 8B Instruct
  • GPT-5 Pro
  • Claude Sonnet 4.5
  • Qwen3 VL 235B A22B Instruct
  • Qwen3 Coder Flash
  • Kimi K2 0905
  • Mistral Medium 3.1
  • GPT-5
  • GPT-5 Nano
  • gpt-oss-20b
  • Codestral 2508
  • Qwen3 30B A3B Instruct 2507
  • GLM 4.5 Air
  • Qwen3 Coder 480B A35B
  • Qwen3 235B A22B Instruct 2507
  • Mistral Small 3.2 24B
  • Gemini 2.5 Pro
  • Gemini 2.5 Pro Preview 06-05
  • Mistral Medium 3
  • Qwen3 14B
  • o3
  • GPT-4.1 Mini
  • Llama 4 Maverick
  • DeepSeek V3 0324
  • Gemma 3 12B
  • Gemma 3 27B
  • Qwen-Plus
  • Command R7B (12-2024)
  • Nova Lite 1.0
  • Nova Pro 1.0
  • Mistral Large 2407
  • Command R (08-2024)
  • GPT-4o (2024-08-06)
  • Llama 3.1 8B Instruct
  • GPT-4o-mini
  • GPT-4o
  • Mixtral 8x22B Instruct
  • GPT-3.5 Turbo (older v0613)
  • GPT-4
See every model and price

Use cases

The work you keep putting off, in one sentence

Each tile is a real run on a demo project: the prompt it was given and the screen it ended on.

Terminal after the prompt: Add tests for the edge cases in cart.js (empty cart, 0% and 100% discount), then run npm test.

Write the tests you skipped

It finds the edge cases, writes the tests and runs them until they pass.

Add tests for the edge cases in cart.js (empty cart, 0% and 100% discount), then run npm test.
Terminal after the prompt: The tests fail. Run node --test, fix the bugs in cart.js, then run the tests again.

Fix the failing build

Say the tests fail. It reads the files, fixes the bugs and reruns the tests.

The tests fail. Run node --test, fix the bugs in cart.js, then run the tests again.
Terminal after the prompt: Refactor cart.js: add a lineTotal(item) helper used by total, keep the exports the same, then run npm test.

Refactor without breaking things

It restructures the code, then runs your tests to show nothing broke.

Refactor cart.js: add a lineTotal(item) helper used by total, keep the exports the same, then run npm test.
Terminal after the prompt: Convert this project from CommonJS to ES modules (package.json type module), then run npm test.

Migrate in one pass

CommonJS to ES modules, one API to the next: it edits every file and checks the result.

Convert this project from CommonJS to ES modules (package.json type module), then run npm test.
Terminal after the prompt: Explain how this project is structured and what each file does. Keep it short.

Learn a codebase you just inherited

Ask how it fits together. It reads the files and explains each one.

Explain how this project is structured and what each file does. Keep it short.
Terminal after the prompt: /review

Review before you push

Reads your uncommitted changes and flags what a reviewer would.

/review
Terminal after the prompt: /security-review

Catch the hole before it ships

Checks your changes for injection, leaked secrets and missing auth checks.

/security-review
Terminal after the prompt: /goal npm test passes and every exported function has a JSDoc comment

Give it the goal, not the steps

It keeps working until the condition is true, then stops.

/goal npm test passes and every exported function has a JSDoc comment
Illustration: a late study session in a libraryIllustration

Inherited a codebase? Ask it anything.

It reads the files and tells you how they fit together.

Illustration: two developers working side by sideIllustration

Subagents

Two jobs. One prompt.

Subagents run at the same time, not one after another

Ask for more than one thing and it starts subagents that work at the same time, each in its own context, then brings their results back together.

  • Subagents run at the same time, not one after another
  • Each has its own context, so your main session stays focused
  • Built-in general and explore agents, or define your own
  • Open a second terminal for another session; both use the same plan
Terminal: two subagents working at once after the prompt "In a single message, call the task tool twice at the same time: subagent 1 writes tests for applyDiscount in discount.test.js; subagent 2 adds JSDoc comments to cart.js. Wait for both, then run npm test."
In a single message, call the task tool twice at the same time: subagent 1 writes tests for applyDiscount in discount.test.js; subagent 2 adds JSDoc comments to cart.js. Wait for both, then run npm test.

Recorded on GPT-6 Luna. Subagents count toward your plan’s limit on requests running at once.

Imagera Code for Chrome

Let it test your site in your own Chrome

A free Chrome extension for Imagera Code. Your session opens your site in a real tab, signed in as you, sees what you see, clicks through it and tells you what is broken. Each session works in its own tab group, only after you allow it.

  • A shop page in a session tab, with a numbered label on every link and button

    It sees the page like you do

    Screenshots with numbered labels on everything it can click.

  • The same shop page shown at phone size, in a phone frame

    Check your site on a phone

    Phone, tablet, laptop and desktop sizes, with touch, in one step.

  • The approval window in Chrome: a coding session asks to pair, with the pairing code pasted in

    Every session asks first

    Allow once, always for a project folder, or decline. The first time, paste the code your terminal shows.

  • The toolbar popup: two connected sessions, each with its last action and a Pause button

    See what each session is doing

    Each session has its own tab group. The toolbar button shows its last action; pause it any time.

  • The toolbar popup asking the person to sign in to the store admin, with a Done button

    Sign-ins and CAPTCHAs stay yours

    It asks you, brings the tab to the front, and waits until you are done.

What it can do

  • Reads the page

    The page as text and structure, any part up close, and screenshots with a number on everything it can click.

  • Uses it like you

    Clicks, types, fills forms, presses keys, scrolls, hovers, drags, and uploads files from your project folder.

  • Finds what is broken

    Console errors, failed requests, slow loading, layout shift, sideways scroll, missing labels and small tap targets.

  • Every screen size

    Phone, tablet, laptop and desktop views with touch, then back to normal in one step.

  • Hands you the hard parts

    Sign-ins, two-step codes and CAPTCHAs: it asks you in Chrome and waits. It never solves a CAPTCHA.

  • You stay in control

    Each session asks first, works only in its own tab group, and can be paused. The extension talks only to Imagera Code on your computer.

Set it up

  1. Add the extension to Chrome

    Free on the Chrome Web Store.

  2. Connect it to Imagera Code

    npx imagera-browser-mcp setup
  3. Ask it to use the browser

    Start imagera-code in your project and ask it to check your site. The first time, Chrome asks you to allow the session: paste the code your terminal shows.

Another MCP client? This prints its config:

npx imagera-browser-mcp setup --print
Illustration: a developer stuck waiting for a limit to resetIllustration: the same developer back in flow

Limits

Stuck at a limit? Keep going.

Every hour stuck at a limit is an hour not shipping.

No short window: you never wait hours for a limit to lift. Past the weekly limit, turn on extra usage and keep going on your API balance or credits, up to a monthly cap you set.

Pro works out to $0.66 a day. Cancel any time. Your plan runs to the end of the period you paid for.

Illustration

Compare

Same starting price. More room to work.

Checked against the published $20 plans of single-lab coding agents, November 2026.

FeatureImagera CodeA $20 single-lab plan
Models 280 from many labsOne lab's models
Short lockout NoneEvery 5 hours
Starting price $20/month$20/month
Illustration: coding at a skate park at golden hourIllustration

Your side project deserves a finish line.

One sentence a day still ships.

Illustration: a team celebrating a launch in a neon arcade barIllustration

Plans

Launch day hits different.

From $20/month. Cancel any time.

One weekly limit, no 3-hour lockouts. Past a limit, keep going on credits: on by default, with a monthly cap you set. Yearly buys the same usage.

Pro

$20/ month

Billed monthly

For short coding sessions and smaller projects.

  • Every coding model, no 3-hour lockouts
  • One weekly limit
  • Keep going on credits when a limit is reached

Max 5x

Most popular

$100/ month

Billed monthly

For daily work across larger codebases.

  • 5× the usage of Pro
  • One weekly limit
  • Keep going on credits when a limit is reached

Max 20x

$200/ month

Billed monthly

For all-day agent work on several projects.

  • 20× the usage of Pro
  • One weekly limit
  • Keep going on credits when a limit is reached

Max 50x

$500/ month

Billed monthly

For teams of agents running in parallel.

  • 50× the usage of Pro
  • One weekly limit
  • Keep going on credits when a limit is reached

Cancel any time; your plan runs to the end of the period you paid for. Taxes may apply.

Illustration: coding on a night trainIllustration

Install

Running in under a minute.

macOS, Linux or Windows. Any terminal.

Three commands on macOS, Linux or Windows, with Node.js and npm. Your key stays on your machine.

  1. Install

    npm i -g imagera-code
    No Node.js? Install without npm

    macOS and Linux

    curl -fsSL https://api.imagera.ai/code/install | bash

    Windows (PowerShell)

    irm https://api.imagera.ai/code/install.ps1 | iex
  2. Sign in

    imagera-code login
  3. Start in any project

    imagera-code

Reference

Models

The list updates by itself as new coding models come out. Prices are credits per million tokens.

280 models
ModelContextInputOutput
GLM 5.3 FlashDefault1M5.039.88
Hy4 flash1M4.4117.7
Qwen4 Max1M180539
Gemini 4 Hydrogen1M189945
ERNIE-Light-DT-1.01M142496
Gemini 4 Carbon1M2521,260
Nex-N2.6262.1K15.150.4
Inkling 2 Small1M37.8126
Claude Fable 5.21M1,0085,040
Muse Ember1M12.663.0
LFM3-1.7B262.1K10.125.2
Mi:dm K 3.0 Pro261.8K25.2101

Pay per use

Pay per use on an Imagera plan

No Code plan? Any paid Imagera plan can use it, paying credits per call with no time limits: nothing resets, nothing to wait for.

In the recording above, fixing both bugs used 0.84 credits on the default model.

FAQ

Questions

Do I need an Imagera Code plan?

A Code plan gives a weekly usage limit with no credits spent, and no 3-hour lockouts. Without one, any paid Imagera plan can use it and pays credits per call.

What happens when I reach a limit?

On a Code plan it keeps going on credits (extra usage) unless you turn that off or reach your monthly cap; then it stops and shows when the limit resets. Turn on auto top-up to refill credits automatically.

Which models can I use?

Every paid text model that can use tools, which coding needs. New ones appear automatically. Pick one with /models inside the app.

What does Imagera keep from my sessions?

For each call Imagera stores the model, token counts and credits, and not your code or prompts. The app sends short crash reports with keys, emails and file paths removed; turn them off with IMAGERA_TELEMETRY=0.

Where do I see my usage?

Settings → Imagera Code shows both windows, spend by model and recent calls.

Can I switch models in the middle of a chat?

Yes. Type /models, pick another one (Claude Opus, Claude Fable, GPT-6 and Gemini and the rest of the list) and keep going in the same session. /effort sets how hard each model thinks.

Does it run on macOS, Linux and Windows?

Yes. Install with npm i -g imagera-code, or without Node.js: curl -fsSL https://api.imagera.ai/code/install | bash on macOS and Linux, irm https://api.imagera.ai/code/install.ps1 | iex in Windows PowerShell. Then sign in with imagera-code login and run imagera-code in a project folder.

Can it run more than one task at once?

Yes. Ask for two things in one message and it starts subagents that work at the same time, each in its own context. A second terminal runs a second session on the same plan.

Can it test my website in a browser?

Yes, with the Imagera Code for Chrome extension. Run npx imagera-browser-mcp setup once; then a session can open your site in your own Chrome, click through it, check it at phone sizes and report errors. Each session asks first and works in its own tab group.

Calling these models from your own app instead of the terminal? Use the LLM API.

Illustration: celebrating a release at sunsetIllustration

Imagera Code

Your next commit is one sentence away.

Pick a plan, install, and hand it the task you have been putting off.

From $20/month: Claude Opus, Claude Fable, GPT-6 and every other coding model.