GenGrades

CometAPI review: pricing, free credits and is it legit?

Published Updated

CometAPI logocometapi.comCreditsWebAPI
The CometAPI homepage: one OpenAI-compatible API for 500+ models.
The CometAPI homepage: one OpenAI-compatible API for 500+ models.Screenshot of cometapi.comtaken

What CometAPI is

In one or two sentences, before anything else.

CometAPI is a prepaid API service that sells text, image, video and audio models through one key at, by its own rule, 80% of the makers' list prices. Its terms and privacy pages name no company and use Hong Kong law, and it says it reaches the low prices through bulk credit buying and routing.

Main features

What it can do, one sentence each, and whether we actually tested it.

  • Official models at 0.8 times list

    CometAPI's pricing page states "Customer price = official x 0.8"; its catalogue lists Claude Opus 5 at Anthropic's 5 / 25 USD with a 0.8 ratio, or 4 / 20 USD per million tokens.

    CometAPI's rule; one docs page still says pay-as-you-go matches official pricing, and Kling endpoints use 0.95.

  • Four request formats

    OpenAI Chat Completions and Responses at api.cometapi.com/v1, Anthropic Messages with prompt caching and server-side web search and fetch, and Gemini's native generateContent.

    Per CometAPI's API reference.

  • Media models without official APIs

    Midjourney (42 catalogue entries) alongside Sora 2, Veo 3.1, Kling, Runway and FLUX 2.

    CometAPI says products without official APIs are implemented "through technical means" and prices them itself.

  • Coding agent guides

    Setup guides for Claude Code, Codex, Gemini CLI, OpenCode, Cline, Continue, avante.nvim, gptme, Hermes Agent, OpenClaw and Pi among 50+ integrations.

    Cursor is not among them.

  • Vercel AI SDK provider and SDKs

    @cometapi/ai-sdk-provider covers text generation, streaming, embeddings and images; there are official Python and Node SDK repositories.

    Per CometAPI's docs and its cometapi-dev GitHub organisation.

  • Per-model status page

    status.cometapi.com probes 56 models and shows 24-hour availability for each.

    Run by CometAPI; its method is not documented.

  • No charge for failed requests

    Gateway rejections and upstream 5xx errors are not billed; completed outputs are billed in full.

    Per CometAPI's refund policy.

Who it is for

The people it helps and the problems it solves, and where it ranks.

Built for

  • Price-driven developers who want Claude, GPT and Gemini below list price
  • Hobby projects mixing chat models with Midjourney, Sora 2 or Kling
  • Claude Code, Codex and Gemini CLI users looking for a cheaper key
  • Developers who want OpenAI, Anthropic and Gemini request formats on one key

Problems it solves

  • Official API prices for frontier models
  • Image tools such as Midjourney that have no official API
  • Separate keys and bills for each model maker
  • Switching coding agents between Claude, GPT and Gemini formats

What it is good at

Public tests other people ran: what they gave it and how it did.

  1. Case 1

    Per CometAPI's Claude Code guide: running Claude Code through CometAPI

    Input

    Set ANTHROPIC_BASE_URL=https://api.cometapi.com and ANTHROPIC_AUTH_TOKEN to a CometAPI key, then start claude with a Claude model ID.

    What happened

    CometAPI accepts Anthropic Messages, so Claude Code needs no other change. By its 0.8 rule Claude Opus 5.5 costs 3.20 / 16 USD per million tokens against Anthropic's 4 / 20 USD; its own status page showed claude-opus-5 at 93.9% availability over 24 hours on October 21, 2026.

  2. Case 2

    In a Hermes Agent GitHub issue (September 25, 2026): calling claude-opus-5-5-thinking through CometAPI

    Input

    A Messages request from Hermes Desktop to model claude-opus-5-5-thinking using the older thinking parameter.

    What happened

    It failed with HTTP 400 passed through from AWS Bedrock Runtime InvokeModel. The error shows the call was served through AWS Bedrock and that CometAPI did not convert the parameter to the adaptive form the model needs. One user's log.

  3. Case 3

    Per CometAPI's AI SDK guide: generating text from a Node.js app

    Input

    With COMETAPI_KEY set: import { cometapi } from '@cometapi/ai-sdk-provider'; generateText({ model: cometapi('your-model-id'), prompt: 'What is CometAPI?' }).

    What happened

    The provider wraps CometAPI's OpenAI-compatible endpoint for generateText, streamText, embeddings and image models. Vendor documentation; no third-party test of the provider has been published.

Where it falls short

What it does badly, with the evidence, and what we could not settle.

Weak

Weak

  • No named company.

    Evidence: Its terms, privacy policy, refund policy, SLA, Trust Center and About page name no legal entity or address; the terms apply Hong Kong law with Hong Kong courts. The domain was registered on July 2, 2024.

  • Model availability below the advertised 99.9%.

    Evidence: Its own status page showed 24-hour availability on October 21, 2026 of 93.9% for claude-opus-5, 95.3% for gpt-5.6 and 97.3% for gpt-6-sol, with a median of 97.9% across 56 probes. The 99.9% SLA covers only the gateway and names no credit.

  • Contradictory and padded price pages.

    Evidence: One docs page says pay-as-you-go matches official pricing and another gives 0.8:1. The catalogue lists 322 models against "500+" in marketing, including unreleased entries such as claude-opus-6 at a placeholder 75 / 75 USD.

  • Sourcing it does not explain.

    Evidence: CometAPI explains its below-list prices only as bulk credit buying, routing and multi-region servers. A user's error shows Claude traffic served through AWS Bedrock, and no page states its arrangements with AWS or the model makers.

  • Rating badges without reviews behind them.

    Evidence: The pricing page shows 5.0 out of 5 on Product Hunt and 4.9 on G2; Product Hunt has no reviews, G2 about two, and Trustpilot none.

  • Not for businesses that must know their supplier or need certified data handling.

    Evidence: With no named operator, "Formal certifications have not yet been initiated", no GDPR section and no list of upstream providers, CometAPI cannot pass a normal vendor review.

Uncertain

Uncertain

  • Whether its Midjourney access is permitted.

    Why unsure: Midjourney has no official API; CometAPI says it is implemented "through technical means" and states no licence.

  • How long 20% below list can last.

    Why unsure: CometAPI gives no reason it can sell under the makers' own prices beyond bulk buying and routing.

  • When the daily maintenance window falls.

    Why unsure: The Help Center names 1 to 5 AM without a time zone.

  • How the service performs under load.

    Why unsure: No third-party latency, throughput or quality test was found; the only per-model data is CometAPI's own status page.

Price, limits and formats

The numbers, each with its source and the date we checked it.

SpecValueNote and source
OperatorNo legal entity named in the terms, privacy, refund, SLA, Trust Center or About pages; Hong Kong law and courts

Source: cometapi.comchecked

Domain registeredcometapi.com, July 2, 2024

Source: rdap.verisign.comchecked

Price vs makers' list prices0.8 times list for official models ("Customer price = official x 0.8"); Kling 0.95

Claude Opus 5 at 4 / 20 USD per 1M tokens against Anthropic's 5 / 25 USD.

Source: cometapi.comchecked

Platform feeNone stated; volume discounts from 3,000 USD of top-ups by arrangement

Source: cometapi.comchecked

Minimum top up and payment10 USD by card through Stripe, the only documented method

Source: apidoc.cometapi.comchecked

Free credit0.1 USD on sign-up, no card required; credits do not expire while the account is in good standing

Source: apidoc.cometapi.comchecked

Models322 in the public catalogue (116 text, 42 video, 40 image, 20 audio); "500+" claimed

Source: api.cometapi.comchecked

API formatsOpenAI Chat Completions and Responses, Anthropic Messages, Gemini generateContent; structured outputs, prompt caching, web search tools

Source: apidoc.cometapi.comchecked

MCPNo official MCP server found in its docs; MCP pass-through not documented

Source: apidoc.cometapi.comchecked

Agent SkillsNo official Agent Skills found in its docs or GitHub organisation

Source: github.comchecked

Coding agentsGuides for Claude Code, Codex, Gemini CLI, OpenCode, Cline, Continue, avante.nvim, gptme, Hermes Agent, OpenClaw, Pi; not Cursor

Source: apidoc.cometapi.comchecked

Privacy and data retentionSays it does not store prompts or outputs or train on request data; logs IP, request frequency and quota use; no certifications

Source: cometapi.comchecked

Status and uptimeGateway SLA 99.9% a month; model probes' 24-hour availability median 97.9%, claude-opus-5 93.9%

Source: status.cometapi.comchecked

RefundsCredits generally non-refundable; unused credits at CometAPI's discretion; disputes within 30 days

Source: cometapi.comchecked

Common questions

Answer first, then why, an example, the limits and what to do next.

Is CometAPI cheaper than the official API?

CometAPI's listed prices are 20% below the makers' own list prices for official models. Claude Opus 5 costs 4 / 20 USD per million tokens against Anthropic's 5 / 25 USD, and GPT-6 Sol 1.60 / 8 USD against OpenAI's 2 / 10 USD.

Limits

The figures apply CometAPI's stated 0.8 rule to its catalogue; one of its docs pages still says pay-as-you-go matches official pricing.

Is CometAPI legit?

CometAPI is a working service with public docs, but it does not say which company runs it. Its terms and privacy pages name no legal entity and choose Hong Kong law, and independent reviews are almost absent: none on Trustpilot or Product Hunt and about two on G2.

Why

Its pricing page still shows 5.0 and 4.9 rating badges for Product Hunt and G2.

Does CometAPI give free credits?

CometAPI gives 0.1 USD of credit on sign-up with no card required. The minimum top-up is 10 USD by card, and credits do not expire while the account is in good standing.

Limits

A third-party guide's figure of about 1 million free tokens is not supported by CometAPI's docs.

Does CometAPI support MCP?

CometAPI does not publish an official MCP server, and none was found in its docs or GitHub organisation. MCP clients such as Claude Code can still use CometAPI as their model provider, but its docs do not describe passing remote MCP servers through its Responses or Messages endpoints.

Is CometAPI reliable?

CometAPI's own status page shows several Claude and GPT models below 98% availability over 24 hours. On October 21, 2026 it showed 93.9% for claude-opus-5, 95.3% for gpt-5.6 and 97.3% for gpt-6-sol, while its 99.9% SLA covers only the gateway.

How does CometAPI compare with OpenRouter?

CometAPI lists lower prices than OpenRouter, while OpenRouter says more about who runs it and lets you choose providers. OpenRouter bills at the provider's list price plus a 5.5% fee on credit purchases; CometAPI charges 0.8 times list, names no company and offers no routing control.

More screenshots

  • CometAPI's pricing page: 20% below official prices for official models, per-call billing for others.
    CometAPI's pricing page: 20% below official prices for official models, per-call billing for others.Screenshot of cometapi.comtaken

What the public tests show

This page is built from CometAPI's own site, docs, terms, status page, public model catalogue and GitHub organisation, the domain registry, the official price pages of OpenAI, Anthropic and Google, GitHub issues, Product Hunt and Hacker News, all checked on October 21, 2026. CometAPI's claims are labelled as its own.

What CometAPI sells

CometAPI sells one prepaid key for text, image, video and audio models from many makers. Its marketing says "500+" models; its public catalogue returned 322 on October 21, 2026 (116 text, 42 video, 40 image, 20 audio, the rest untyped). They include GPT-6 Sol, Claude Opus 5.5, Gemini 4 Argon, Grok 4.7, DeepSeek V4.1, Kimi K3, Sora 2, Veo 3.1, Kling, Runway, FLUX 2 and 42 Midjourney entries. The catalogue also lists models the makers have not released, such as claude-opus-6, marked "Coming Soon" and priced at a placeholder 75 / 75 USD.

New accounts get 0.1 USD of credit with no card. Top-ups start at 10 USD by card through Stripe, the only documented payment method, and credits do not expire while the account is in good standing. When an upstream channel fails, CometAPI's Help Center says the system switches channels and retries; callers cannot choose which upstream serves a request.

Prices against the official APIs

CometAPI's pricing page states the rule "Customer price = official x 0.8" for official models (Kling endpoints use 0.95). Its catalogue lists each model's official price and a 0.8 ratio. The CometAPI column below applies that rule to the catalogue:

Model, per million input / output tokensCometAPIMaker's list priceSource
Claude Opus 5.53.20 / 16 USD4 / 20 USDAnthropic
Claude Opus 54 / 20 USD5 / 25 USDAnthropic
Claude Sonnet 5.51.60 / 8 USD2 / 10 USDAnthropic
GPT-6 Sol, up to 272K input tokens1.60 / 8 USD2 / 10 USDOpenAI
Gemini 4 Argon1.60 / 8 USD2 / 10 USD until November 30, 2026; 4 / 20 USD from December 1Google

CometAPI explains the discount as an "intelligent routing system", buying API credits in bulk from several providers and servers in several regions. Its docs do not agree with themselves: one page says pay-as-you-go is "consistent with official pricing", another gives the 0.8:1 ratio, and the Help Center describes per-model "multipliers" that do not map directly to official prices and change with market conditions for non-GPT models. Midjourney, Kling, Luma and other products without official APIs are priced by CometAPI itself, which says it implements them "through technical means".

What public tests show

No third-party latency, throughput or output-quality test of CometAPI was found in public sources. What exists:

  • CometAPI's own model status page (Uptime Kuma probes of 56 models, live on October 21, 2026) showed 24-hour availability of 93.9% for claude-opus-5, 94.5% for claude-opus-4-7, 95.3% for gpt-5.6 and 97.3% for gpt-6-sol. Gemini and DeepSeek probes were at 98% to 99%, the median was 97.9% and the lowest 60.6% (grok-build-0.1). The method is not documented.
  • A GitHub user in the Hermes Agent tracker (September 25, 2026) called claude-opus-5-5-thinking through CometAPI and got an HTTP 400 passed through from AWS Bedrock Runtime InvokeModel. The log shows the request was served through AWS Bedrock and that CometAPI did not convert the old thinking parameter to the adaptive form the model requires.
  • CometAPI's own latency benchmark page defines time to first token and throughput; its sample figures (640 ms, 42.1 tokens a second) are labelled as an illustrative format, not measured CometAPI results.
  • A CometAPI blog post compares CometAPI with OpenRouter and AIML API; it is a competitor comparison written by the vendor and contains no measurements.

What users report

Independent reviews barely exist. Trustpilot has no reviews for cometapi.com (the unclaimed profile); negative reviews that turn up for "Comet" are about cometgift.com, an unrelated retailer. Product Hunt shows 6 points, 12 followers, no reviews and one comment, from the maker. G2 shows about two reviews in search results. CometAPI's pricing page nevertheless carries badges of 5.0 out of 5 on Product Hunt and 4.9 out of 5 on G2.

Most visible activity is promotion. On Hacker News, an account named cometapi posted 61 items and other accounts posted identical marketing text. On GitHub, an account opened issues on more than 20 open-source projects in September and October 2025 (the Vercel AI SDK, Aider, Milvus and others) asking them to add CometAPI as a provider, and an account named CometAPI-Official continues in 2026. A press release on GPT-6 Astra went out through GetNews in September 2026.

MCP, Agent Skills and coding agents

No official MCP server was found in CometAPI's docs or on its GitHub organisation (cometapi-dev). Its Responses API docs cover function calling and the built-in web_search tool and refer to OpenAI's tools guide for other built-in tools; passing a remote MCP server through a request is not documented. Its Messages API docs cover prompt caching and the server tools web_search and web_fetch, but no MCP connector and no Anthropic Skills. No official Agent Skills were found either.

For coding agents, the docs list more than 50 integration guides, including Claude Code (set ANTHROPIC_BASE_URL=https://api.cometapi.com and ANTHROPIC_AUTH_TOKEN), Codex, Gemini CLI, OpenCode, Cline, Continue, avante.nvim, gptme, Hermes Agent, OpenClaw and Pi. Cursor is not among them. A desktop helper, CometAPI Connect, writes CometAPI settings into installed tools; its README said the v0.1.3 release was private while CometAPI completes a publication review. For code, there is a Vercel AI SDK provider (@cometapi/ai-sdk-provider), official Python and Node SDK repositories, guides for LiteLLM, LangChain and LlamaIndex, and an llms.txt index of the docs.

Privacy, terms and who runs it

CometAPI's terms, privacy policy, refund policy, SLA, Trust Center and About page name no legal entity and no address; the footer carries only a CometAPI copyright line. The terms choose Hong Kong law with exclusive jurisdiction of the Hong Kong courts. The domain cometapi.com was registered on July 2, 2024, and its first Hacker News posts date from January 2025. The Help Center uses the vocabulary of the open-source One API and New API relay software: channels with automatic switching, group and completion multipliers, and a Tokens menu for API keys.

  • CometAPI says it does not collect or store prompts, inputs or outputs, and records IP addresses, request frequency and quota use. The Trust Center says request data is not used for training and "Formal certifications have not yet been initiated"; the About page calls SOC 2 Type II a goal.
  • There is no GDPR section, data region, list of upstream providers or named data controller.
  • Credits are generally non-refundable; unused credits may be refunded at CometAPI's discretion, and disputes must be raised within 30 days. Gateway rejections and upstream 5xx errors are not billed.
  • The SLA targets 99.9% monthly uptime for the gateway only, excludes upstream disruptions and names no service credit. A daily maintenance window runs from 1 to 5 AM, time zone not stated.

How it compares with other relays

ServicePrice against makers' list pricesWho runs it, per its own terms
CometAPI0.8 timesNo company named; Hong Kong law
OpenRouterSame, plus 5.5% on credit purchasesOpenRouter, Inc.
AIMLAPIAbout 1.3 timesBoiler Labs FZ-LLC; AIMLAPI OÜ
ZenMuxSame, 5% top-up fee waived for nowNexaMind Singapore Pte. Ltd.
Kie.aiBelow list for many modelsTwo LLCs at a Denver address

CometAPI is the cheapest of these for Claude, GPT and Gemini on paper. The others either charge list price or more, but say who you are contracting with, and OpenRouter lets callers choose providers.

Who should skip it

Businesses that must know their supplier, need certified data handling or data-residency terms, or need steady Claude and GPT access for production work. For personal projects where a 20% saving matters more than an anonymous operator, the risk is bounded by the credit you top up.

Sources

Sources

  1. CometAPI homepage (CometAPI) (cometapi.com), accessed
  2. CometAPI pricing page: the 0.8 rule and rating badges (CometAPI) (cometapi.com), accessed
  3. CometAPI public model catalogue API (CometAPI) (api.cometapi.com), accessed
  4. CometAPI terms of service: Hong Kong law (CometAPI) (cometapi.com), accessed
  5. CometAPI privacy policy (CometAPI) (cometapi.com), accessed
  6. CometAPI refund policy (CometAPI) (cometapi.com), accessed
  7. CometAPI SLA (CometAPI) (cometapi.com), accessed
  8. CometAPI Trust Center (CometAPI) (cometapi.com), accessed
  9. About CometAPI (CometAPI) (cometapi.com), accessed
  10. CometAPI docs index, llms.txt (CometAPI) (apidoc.cometapi.com), accessed
  11. CometAPI important guidelines: free credit, top-up, payment (CometAPI docs) (apidoc.cometapi.com), accessed
  12. CometAPI Help Center: multipliers, channels, maintenance (CometAPI docs) (apidoc.cometapi.com), accessed
  13. CometAPI text and chat APIs overview (CometAPI docs) (apidoc.cometapi.com), accessed
  14. CometAPI Responses API reference (CometAPI docs) (apidoc.cometapi.com), accessed
  15. CometAPI Anthropic Messages reference (CometAPI docs) (apidoc.cometapi.com), accessed
  16. CometAPI integrations index (CometAPI docs) (apidoc.cometapi.com), accessed
  17. Use Claude Code with CometAPI (CometAPI docs) (apidoc.cometapi.com), accessed
  18. Use AI SDK with CometAPI (CometAPI docs) (apidoc.cometapi.com), accessed
  19. cometapi-dev GitHub organisation: SDKs, AI SDK provider, CometAPI Connect (github.com), accessed
  20. CometAPI model status page (Uptime Kuma, run by CometAPI) (status.cometapi.com), accessed
  21. AI API latency benchmark methodology (CometAPI blog) (cometapi.com), accessed
  22. CometAPI vs OpenRouter vs AIML API, a competitor comparison by CometAPI (CometAPI blog) (cometapi.com), accessed
  23. cometapi.com domain registration (Verisign RDAP) (rdap.verisign.com), accessed
  24. Claude via CometAPI served through AWS Bedrock (Hermes Agent GitHub issue #122672, September 25, 2026) (github.com), accessed
  25. Request to add CometAPI as a provider (Vercel AI SDK GitHub issue #9001) (github.com), accessed
  26. cometapi.com profile, no reviews (Trustpilot) (trustpilot.com), accessed
  27. CometAPI on Product Hunt (Product Hunt) (producthunt.com), accessed
  28. Hacker News search results for cometapi (HN Algolia) (hn.algolia.com), accessed
  29. OpenAI API pricing: GPT-6 Sol (OpenAI) (developers.openai.com), accessed
  30. Claude API pricing: Opus 5.5, Opus 5, Sonnet 5.5 (Anthropic) (platform.claude.com), accessed
  31. Gemini API pricing: Gemini 4 Argon (Google) (ai.google.dev), accessed
  32. OpenRouter pricing: list prices and 5.5% credit fee (OpenRouter) (openrouter.ai), accessed
  33. OpenRouter terms of service: OpenRouter, Inc. (OpenRouter) (openrouter.ai), accessed
  34. About Kie.ai: its two operating LLCs (Kie.ai) (kie.ai), accessed

Changelog

Changelog

  1. Page created from CometAPI's own pages, public tests and user reports.

Recommended from GenGrades

  • Tool review

    APIMODELS

    APIMODELS is a pay-as-you-go AI API relay that sells about 150 models through one key at its own prices, mostly below the makers' list prices, and spends the margin on backup channels: 98.62% of requests succeeded over 30 days, moderation rejections excluded. APIMODELS is our own product: it is made by the team behind GenGrades.

  • Tool review

    ZenMux

    ZenMux is an LLM gateway, operated under its terms by NexaMind Singapore Pte. Ltd., that sells Claude, GPT, Gemini and many Chinese models through OpenAI, Anthropic and Vertex-style endpoints at the makers' list token prices. It adds automatic credits for bad or slow answers, which it calls insurance, and 20 to 200 USD monthly coding plans.

  • Tool review

    WaveSpeedAI

    WaveSpeedAI is an API platform, run by a Singapore company and a Hong Kong company named in its terms, that sells 1,000+ image, video, audio and 3D models plus an LLM service on one balance. It matches fal on Kling 3.0, charges above Google's list for Veo 3.1 Fast and accepts Alipay and WeChat Pay.

  • Tool review

    Vercel AI Gateway

    Vercel AI Gateway is Vercel's hosted API for about 480 text, image, video and speech models behind one key, billed from prepaid credits at the providers' list prices with no markup, including when you bring your own keys. It became generally available on August 21, 2025, and gives each Vercel team 5 USD of free credit a month.

  • Tool review

    Requesty

    Requesty is a London-based LLM gateway, run by Requesty Ltd and seed-funded by 20VC in 2025, that sells OpenAI- and Anthropic-compatible access to more than 600 models at list price plus a 5% top-up fee. It stands out for routing policies, spend controls and an EU endpoint in Frankfurt.

  • Tool review

    Replicate

    Replicate is an API for running AI models, owned by Cloudflare since a $57.4 million acquisition closed on December 1, 2025. It sells 100+ official models such as Veo 3.1, Nano Banana Pro and Kling 3.0 per output, and 50,000+ community models by the second of GPU time.