Scout Forge
Usability74Design72Speed88Security40Accessibility80Growth58

Cycling through all six. Tap any point to stop.

Measured on six things

One API key for Claude Code, Codex and Gemini — billed per token, from 1/6 of the official price.

Security

40

HTTPS enforced but no HSTS, CSP, frame protection, X-Content-Type-Options, Referrer-Policy or Permissions-Policy; SPF missing. Checklist 3/13. For a service that holds your API keys and billing, that's a thin security posture.

Screenshots

1 / 1

About this app

Use Claude Code, Codex and Gemini CLI with one API key and one endpoint. Pay per token from 1/6 of official prices across 21 models, no subscription.

Key features

  • 1One API key and one prepaid balance across Claude, GPT and Gemini model families
  • 2Four wire formats: Anthropic Messages, OpenAI Responses, OpenAI Chat Completions, Gemini generateContent
  • 3Per-token pricing at 1/7.5 (Claude, Codex) to 1/10 (Gemini) of official list price, no subscription
  • 4One-command installer for Claude Code, Codex, Gemini CLI and 9 more agents, with --uninstall support

Target customer profile

Drawn from the product itself, not from a survey.

Demographic

Solo developers & indie hackers running AI coding agents daily

Pain points

Official Claude/GPT/Gemini API bills balloon fast; multiple subscriptions are messy

Primary needs

One cheap key that works with Claude Code, Codex and Gemini CLI

Demographic

Small dev teams & startups with tight tooling budgets

Pain points

Separate sign-ups, payment methods and invoices per vendor; no unified spend view

Primary needs

Centralized billing, prepaid top-ups and volume discounts

Demographic

Engineering teams multi-tasking across model families

Pain points

Rewiring SDKs and agents every time they switch between Claude, GPT and Gemini

Primary needs

One base URL with Anthropic, OpenAI and Gemini API format compatibility

Demographic

Students and hobbyist tinkerers learning agentic coding

Pain points

Free tiers and official pricing make experimentation expensive

Primary needs

No-card signup, pay-per-token prepaid access and step-by-step setup guides

Scout Review

Written by AI from measured evidence, scored out of 100.

Code By AI is a multi-model API reseller aimed squarely at people running Claude Code, Codex and Gemini CLI all day. The pitch is clean: one key, one balance, four wire formats, 1/7.5–1/10 of list price, no subscription. The execution on the marketing side is above average — fast site, deep docs, a real installer, a pricing page that shows its work. The measured security posture is the weak spot: 3/13 on headers for a service that holds your API keys. And the model IDs on the quickstart don't match anything publicly verifiable, which is the kind of detail that matters when the whole product is 'trust us with your tokens.'

Measure by measure

Usability

74

13 agent setup guides, a one-command installer that backs up configs and supports --uninstall, and a worked pricing example. Onboarding is genuinely thought through. No free trial tokens and no live demo hold it back.

Design

72

Clean dark/light theme, consistent brand, terminal-style code blocks and a genuinely useful pricing comparison table. Dense but organized. Held back by unverifiable model IDs in the quickstart and a very long single-scroll homepage.

Speed

88

Lighthouse mobile 97/100 (LCP 1.2s, CLS 0.003, TBT 181ms) and desktop 99/100. The site is content-heavy but still snappy, pointing to static/edge delivery. No user complaints about lag found.

Security

40

HTTPS enforced but no HSTS, CSP, frame protection, X-Content-Type-Options, Referrer-Policy or Permissions-Policy; SPF missing. Checklist 3/13. For a service that holds your API keys and billing, that's a thin security posture.

Accessibility

80

Lighthouse accessibility 96/100 with a language switcher and theme toggle. Two real failures: insufficient color contrast and SVGs with img role missing text alternatives. No public WCAG statement.

Growth

58

Real problem (three API bills, three keys), sharp wedge (one key, four protocols, 1/7.5–1/10 list price), specific audience. Domain is <1yr old. Durability of the resale margin is the open question.

Improvement areas

  • Security headers: 3/13 on the HTTP probe — no HSTS, CSP, frame protection, Referrer-Policy or SPF
  • Model IDs on the quickstart (claude-sonnet-5, gpt-5.6-terra) don't match verifiable public catalogs
  • No free trial credit — signup is free but you must top up before you can test the endpoint
  • No public security page, SOC 2, pen-test or bug-bounty disclosure for a key-holding gateway

Design & Usability

The site is a competent, modern dark/light build with a consistent brand, terminal-style code blocks and a pricing page that actually does the math for you (20M in / 2M out, side by side against list price). Usability is the standout: 13 agent setup guides covering Claude Code, Codex, Gemini CLI, Cline, Trae, Qwen Code and more, plus a one-command installer that backs up your existing configs and supports --uninstall. That's more onboarding care than most resellers bother with. The drag is trust-by-detail: the quickstart advertises model IDs like claude-sonnet-5 and gpt-5.6-terra that don't map to anything I can verify, which undercuts the polish. No free trial tokens either — you sign up free, then top up before you can test anything.

Speed & Security

Speed is the strongest measured dimension: Lighthouse mobile 97/100 with LCP 1.2s and CLS 0.003, desktop 99/100. The site is content-dense but still fast, which points to static or edge delivery. Security is the opposite story. The HTTP probe found HTTPS enforced but no HSTS, no CSP, no frame protection, no X-Content-Type-Options, no Referrer-Policy, no Permissions-Policy, no secure cookies and no SPF — 3 of 13 checklist points. For a service whose entire job is holding your Anthropic, OpenAI and Gemini API keys and your prepaid balance, that's a real gap, not a missing-signal technicality. No SOC 2, ISO or bug-bounty page surfaced in the fetched content.

Accessibility & Growth

Accessibility scores 96/100 on Lighthouse with two concrete failures: insufficient color contrast and SVG elements with an img role lacking accessible text alternatives. There's a language switcher, which suggests i18n intent, but no WCAG statement. Growth is a judgement on the idea: the problem is real (three vendors, three bills, three keys), the wedge is specific (one key, four wire protocols, 1/7.5–1/10 of list price, no subscription), and the audience is named. Against OpenRouter the pitch is cheaper and more agent-focused; against the official APIs it's a straight discount. The honest limit is that reselling frontier tokens at a tenth of list is a margin question, not a product question — the domain is under a year old and the model names on the site can't be verified against public catalogs.

Conclusion

If you're burning real money on three separate API bills and you want one key and one balance, this is a coherent answer with better docs than most of the category. Go in with eyes open: top up small first, check the model IDs against the vendor catalogs yourself, and don't route anything you can't afford to lose through a gateway that hasn't shipped an HSTS header. The idea is good. The trust layer needs to catch up to the landing page.

Suggestions for improvement

  • Ship the standard security headers — HSTS, CSP, X-Content-Type-Options, Referrer-Policy, Permissions-Policy — and add SPF. 3/13 is not a look for a key-holding gateway.
  • Publish a security page: data handling, key storage, logging policy, and ideally a third-party pen test. Resellers live or die on this.
  • Reconcile the model IDs on the quickstart with public vendor catalogs, or add a note explaining the naming. Unverifiable model names erode trust fast.
  • Offer a small free trial credit (even $1) so people can validate the endpoint before topping up. 'No card to sign up' is good; 'no way to test' is not.
How we score

Alternatives comparison

Named competitors, point by point. Nobody paid to appear here or to be left out.

Code By AIOpenRouterAnthropic API (official)OpenAI Platform (official)Google AI Studio / Gemini API (official)
Pricing modelPay-per-token at 1/7.5–1/10 of list price, prepaid top-up, no subscriptionPay-per-token at or near provider list price, credit-based, no subscriptionOfficial list price per token, postpaid billingOfficial list price per token, prepaid creditsOfficial list price, generous free tier for Gemini
Model coverage21 models across Claude, GPT and Gemini familiesHundreds of models across dozens of providersClaude models onlyGPT and OpenAI models onlyGemini models only
API format compatibilityAnthropic Messages, OpenAI Responses, OpenAI Chat Completions, Gemini generateContentPrimarily OpenAI-compatible; Anthropic-style via adaptersAnthropic Messages onlyOpenAI Responses and Chat Completions onlyGemini generateContent only
Coding-agent setup guides13 guides incl. Claude Code, Codex, Gemini CLI, Cline, Trae, Qwen Code, plus one-command installerDocs cover SDKs and some agent integrations; no dedicated per-agent installerClaude Code setup documented; other agents not coveredCodex CLI setup documented; other agents not coveredGemini CLI setup documented; other agents not covered
Security & compliance signalsHTTPS only; 3/13 security headers; no SOC 2/ISO/pen-test page foundHTTPS, standard headers, published privacy policy; SOC 2 not verified hereEnterprise-grade: SOC 2, ISO, published trust centerEnterprise-grade: SOC 2, ISO, published trust centerEnterprise-grade: Google Cloud compliance stack

OpenRouter

A popular multi-model API aggregator with one key and unified billing across many providers, often used for model routing and fallbacks.

Anthropic API (official)

Direct official access to Claude models for Claude Code and the Anthropic SDK, with full support and standard list pricing.

OpenAI Platform (official)

Official access to GPT models and the Responses/Chat Completions APIs used by Codex CLI and the OpenAI SDK.

Google AI Studio / Gemini API (official)

Google's official Gemini API and AI Studio for generateContent endpoints, used by Gemini CLI and the Google SDK.

Comparing options? See Code By AI alternatives, scored side by side

More Development & Engineering apps

View all Development & Engineering
Also in:API ServiceProductivityAI Generations
Scout Forge

A platform dedicated to providing unbiased reviews of newly launched applications, analyzing everything from their features to their full potential.

info@scoutforge.net

© 2026 Scoutforge. All rights reserved.

Akatron NetworkCrafted by Akatron Network