R的API小站 offers API services for accessing various AI models, primarily from Google and Anthropic. Key models include Gemini-3-Flash (noted for speed and cost-effectiveness), Gemini-3-Pro (slower but with strong conversational abilities), and Claude models (Opus, Sonnet, Haiku). The provider emphasizes transparency in model availability and quick issue resolution for commonly used models. Use cases include AI-powered applications, chatbots, and content generation. No specific pricing, SLAs, or support details are provided in the content.
R的API小站
api.xiaor.online
Provides API access to AI models including Gemini and Claude, with a focus on speed and reliability.
- Models
- 135 models
- From
- $0.230/M
- Speed
- 74 tok/s
- Updated
- 6/25/2026
- Latency
- 0.00 s
- Created At
- 2/21/2026
- Recharge Rate
- ¥0.23 per $1 quota
Features
Login Methods
API Endpoints
- Original Route
https://xiaor.xx.kgDirect server connection with higher latency in China
- Endpoint 2Historical / Unverified
https://ali.xiaor.xx.kg - CF Optimized
https://api.xiaor.onlineRecommended, but may have issues with poor proxy connections
- Ali Acceleration Domestic Line
https://api.lrca.cnTop-tier
- Fastest Route
https://xiaor.yd8.topFast but less stable compared to CF Optimized
- Recommended: Tencent Global Acceleration
https://rapi.yd8.topStrongly recommended for mainland users
Verify ownership to unlock provider management features:
- Edit provider name, content, and links
- Get featured with priority traffic and visibility boost
- Display a verified badge to build user trust
Leaderboard Rankings
Rankings are based on community-submitted tests and periodic health probes. Advisory only, not official data.
About R的API小站
Health Check
API Benchmarks & Pricing
Compare 231 model rows across audit recency, latest speed tests, throughput, latency, and per-token pricing.
[AG-F8][量]gemini-3.6-flashdefault | $2.02/M | $9.20/M |
[AG-F9]gemini-3.1-pro-previewdefault | $2.76/M | $16.56/M |
[AG-F9]gemini-3.5-flash-highdefault | $1.84/M | $11.04/M |
[AG]gemini-3.1-pro-lowdefault | $0.110/request | - |
[AG]gemini-3.1-pro-low纯自建 | $0.083/request | - |
[AG]gemini-3.1-pro-low-流式抗截断default | $0.184/request | - |
[AG]gemini-3.1-pro-low-流式抗截断纯自建 | $0.138/request | - |
[AG]gemini-3.6-flash-tiered纯自建 | $1.38/M | $8.28/M |
[AG]gemini-3.6-flash-tiereddefault | $1.84/M | $11.04/M |
[C-F2][次]gemini-2.5-prodefault | $0.092/request | - |
[C-F2][次]gemini-2.5-pro-searchdefault | $0.092/request | - |
[C-F2][次]gemini-3-flash-previewdefault | $0.020/request | - |
[C-F2][次]gemini-3.1-pro-previewdefault | $0.129/request | - |
[C-F2][次]gemini-3.1-pro-preview-maxthinking-searchdefault | $0.138/request | - |
[C-F2][次]gemini-3.1-pro-preview-searchdefault | $0.092/request | - |
[C]gemini-3.5-flash-extra-low-流式抗截断default | $0.156/request | - |
[C]gemini-3.5-flash-extra-low-流式抗截断纯自建 | $0.117/request | - |
[C]gemini-3.5-flash-low纯自建 | $0.048/request | - |
[C]gemini-3.5-flash-lowdefault | $0.064/request | - |
[F][次]grok-4.5default | $0.092/request | - |
[F][量]grok-4.5default | $0.920/M | $2.76/M |
[Kiro-F3][次]claude-opus-4-8default | $0.294/request | - |
[Kiro-F3][次]claude-opus-5default | $0.294/request | - |
[Kiro-F3][次]claude-sonnet-5default | $0.147/request | - |
[OL]gemma4:31b纯自建 | $0.028/request | - |
Showing 25 of 231 model rows
Recent Test Records
| Time | Model | Speed | Latency |
|---|---|---|---|
| Feb 21, 03:52 PM | [AG-2][0.2/次]gemini-3.1-pro-high | 73.88 tok/s | 14.28s |
Similar API Provider Alternatives to Compare
Compare R的API小站 alternatives against 6 nearby API providers using 243 LMSpeed signals across shared model coverage, pricing, benchmark speed, uptime, and free-model availability.
| Provider | Why compare | Models | Free | Avg price | Speed | 30d uptime |
|---|---|---|---|---|---|---|
| R的API小站 api-xiaor-online Provides API access to AI models including Gemini and Claude, with a focus on speed and reliability. | Current provider baseline | 135 | 0 | $0.345/M | 74 tok/s | 9940% |
| 钱多多 API api2-aigcbest-top Provides AI-generated content APIs for various applications, including text and image generation. |
| 1,418 | 2 | $0.0025/M | 101 tok/s | 9940% |
| F2API api-f2api-com F2API is an enterprise AI gateway providing a unified OpenAI-compatible interface for accessing multiple large language models. |
| 1,262 | 0 | N/A | 272 tok/s | 9990% |
| Koyeb AI Gateway new-api-koyeb-app An OpenAI-compatible API gateway deployed on Koyeb, providing access to multiple AI models. |
| 578 | 420 | N/A | 33 tok/s | 9960% |
| 哈基米API站 api-gemai-cc 哈基米API站 offers a unified API gateway for large language models with competitive pricing and stable service. |
| 323 | 0 | $0.930/M | 119 tok/s | 9980% |
| TommyLam API new-api-tommylam-me TommyLam API offers an OpenAI-compatible API gateway for accessing multiple AI models at competitive rates. |
| 231 | 100 | N/A | 61 tok/s | 10000% |
| SMLC666 API api-smlc666-top SMLC666 API provides an OpenAI-compatible API relay service with unified access to various AI models. |
| 183 | 0 | $0.240/M | 43 tok/s | 9970% |
Announcements
If payment goes through with no response, please contact the admin. Also, temporarily ignore the "Legacy frontend will be discontinued soon" notice; the new UI is expected to be officially deployed on July 10 (originally scheduled for July 20).
Celebrating the first anniversary of the site! The 001 plan + 1:5 top-up event is now live. A lucky draw will be available later too.
The DeepSeek V4 Pro cache for pay-per-use has some issues. Actual responses are fine, but the site seems unable to detect cached responses. So until fixed, input price is reduced to 1 credit/M, output and cache prices unchanged. Also, later we will compensate users with 80% of the consumption credits from the time the issue was reported until the price reduction. That is all.
CC-F3 temporarily removed. Will be re-added after adding R18 content detection. Do not use this channel for tasks violating the A\ agreement.
Check-in credits changed to 0~2, and the check-in mechanism has been slightly modified.
gpt-image-2 must be called via the /v1/image path, otherwise it will report "No available channel".
Kiro's Claude Opus member accounts are almost all gone...
On 2026.6.1T00:00:00, accounts with zero calls and zero payments in the past 30 days (excluding 001 plan benefits) will be deactivated. Also, check-in pool credits for accounts with zero calls in the past 7 days and full check-in credits will be cleared. Please be aware of your usage.
Bad news: According to AG shunt 3 tests, anti-gravity now shares Gemini model quotas, no longer distinguishing between pro and flash. Therefore, flash price is raised to the Pro model price, i.e., 0.1 credits per call.
Self-built Kiro channel for Claude is now live, 0.12 credits per call.
Gemini 3.5 Flash is now available via self-built and shunt channels, channel AG anti-gravity, 0.02 credits per call. Welcome to try.
You can try [Thick Armor] Claude, 0.3 credits/call. But this armor is easier to break through than Kiro. If many people use it, the price can be reduced to 0.2 credits/call.
After May 1, unused total check-in credits will be limited. You can check the relevant cap and current credit status on the "Activity Lottery" page on the homepage. Simply put, there is a check-in reservoir. When it's full and the water (credits) is not consumed, continued check-ins trigger a mechanism, and the excess will be deducted as "management" message level.
GPT supply restored (now as relay). Please use the half-price group to access GPT closer to the intended pricing. Reference GPT-5.4 real pricing:
Input price ¥0.3130 / 1M Tokens Completion price ¥1.8750 / 1M Tokens Cache read price ¥0.0750 / 1M Tokens
AS channel is completely down, please switch to other channels to use
About AG Xiao Ke: Due to the impact of Google AG quota adjustment (U's quota reduced to 1/6), which drives changes in reverse proxy costs, prices will be adjusted on the morning of 03.25. Specific adjustments will be seen later, but first, an announcement to sync information. Sigh, it's estimated to double.
About CC Xiao Ke: When calculating AG Xiao Ke's accounts, I realized that CC Xiao Ke with discounts is at a loss... Therefore, on April 2nd, CC Opus input will be increased from 30 quota to 33, output adjusted proportionally. After adjustment, it will break even, and it's more recommended to use it when there are discounts (pray).
Shimizu (Kiro)'s [self-built] and [diversion] are almost all blown up... The free account usable 4.5S could originally be set to 0.2x, but now that it's blown up, we can only take it down.
Please note that the Shimizu channel is currently unavailable, calls will show errors. It's not excluded that in the future, the channel might be temporarily disabled, leading to displays like XX group unavailable XX model, so mentioning it here.
The specific price adjustment plan is as follows:
- 1️⃣[Shimizu-diversion] will be removed from the half-price group on March 22, 2026, temporarily retaining 75% discount
- 2️⃣[AG] and its diversion's Claude-Opus and Claude-Sonnet per call will be increased from the original 0.3 quota/call (0.25 quota/call) to 0.5 quota/call (0.45 quota/call), supporting stacking with 75% discount group.
Sigh...
Sigh, after CC price increase, Shimizu Kiro increases. Self-built might not be maintained for a while because it's too harsh, and the diversion price quadrupling is logical... Also, AG Xiao Ke's quota has become 1/6 of the original... AG also has to increase...
FAQ
How do I configure the API for this website?
You can refer to the content on the second screen of the homepage, which has a 1:1 explanation. **Choose OpenAI**: Additionally, whether you are calling Gemini or Claude, make sure to select OpenAI from the three API interface options: OpenAI, Gemini, Claude. Because this site aggregates AI models from various sources, converting these different sources into an OpenAI interface for unified API calls, so when you see these three options, choose OpenAI. **Check if the API URL and Key are filled correctly**: Some software like Rikkahub has the 'API URL' in the upper box and the KEY in the lower box, so please verify.
Is it necessary to use SillyTavern to interact with AI?
Of course not. In fact, you can try using Rikkahub, Kelivo, or CherryStudio as platforms for interacting with AI. These platforms support custom APIs, can connect to this site's AI models, and are more lightweight than SillyTavern, suitable for Q&A, organizing materials, AI teaching, and other tasks. Additionally, Rikkahub currently only has an Android version, Kelivo supports Windows and Android, and CherryStudio supports all platforms including Apple devices (but on Android, the first two are still more visually appealing and user-friendly).
Notes
- Health checks: Scope: the 72-hour chart and recent availability measure API connectivity only. Each bar summarizes one hour of checks. Targets: LMSpeed tries the configured health check URL and provider status URL first, then API endpoints derived from known API hosts and recent speed-test base URLs. A website host is considered only when it looks like an API endpoint. Probe steps: each candidate goes through DNS lookup, TCP connection, TLS handshake for HTTPS, and an HTTP HEAD request with redirects followed. Probing stops after the first reachable candidate. Reachable criteria: every required network step must succeed. An HTTP response below 500 is treated as reachable, including 401 because it confirms that an authenticated API endpoint responded, except for statuses classified as blocked. Blocked results: HTTP 403, 429, 521, 525, and 530, plus detected WAF or Cloudflare challenges, are shown as blocked and excluded from availability calculations because LMSpeed cannot determine whether the API itself is down. Model availability: when a dedicated test key is configured, LMSpeed sends an authenticated GET request to a derived /models endpoint and compares returned model IDs with this provider's listed models. These per-model results appear in Models & Pricing and are not included in the provider connectivity percentage. Timeouts: TCP connection, TLS handshake, HTTP connectivity, and model requests each use a 20-second timeout. A full run can take longer when several candidates are tried. Frequency: a background worker checks all providers every 5 minutes by default. The 72-hour chart combines those samples into hourly bars, and the schedule may be changed by the service operator. Limit: automated samples are not an SLA and do not guarantee account quota, every model, every region, or successful completion requests. Check the provider's own status page before making operational decisions.
- Domain Rating data is sourced from Ahrefs. It is a 0–100 backlink-based domain strength signal and does not measure API speed or reliability.
- Announcements and FAQ are read from this provider's NewAPI status snapshot when available. LMSpeed stores the original content and optional English translations from the provider status source, then shows the localized fields on this page.

