VseGPT alternative — the latest Claude, GPT and Gemini about 3x cheaper
RuAPI is a VseGPT alternative with no subscription fee: you pay only for tokens, and input and output for Claude Sonnet / Opus 5.5, GPT-5.4 and Gemini 3.1 Pro cost 2.6–3.1 times less (vsegpt.ru price list on 01.11.2026, Bank of Russia rate). A native Anthropic API with prompt caching for Claude Code, and the Responses API for Codex. Top up in USDT from 5 USDT — no foreign card and no VPN needed.
from openai import OpenAI
client = OpenAI(
api_key="sk-...",
base_url="https://www.ruapi.ai/v1",
)
resp = client.chat.completions.create(
model="gpt-5.4",
messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)OpenAI SDK code moves over as is: replace https://api.vsegpt.ru/v1 with https://www.ruapi.ai/v1, swap the VseGPT key for a RuAPI key and drop the prefix from the model ID: openai/gpt-5.4 → gpt-5.4.
Why people look for a VseGPT alternative
The latest Claude, GPT and Gemini — about 3x cheaper.
Claude Sonnet 5.5 costs $2.40 / $12.00 per 1M input / output tokens on RuAPI vs 0.6 ₽ / 3.0 ₽ per 1,000 on VseGPT (≈ $7.45 / $37.24 per 1M). GPT-5.4: $3.00 / $18.00 vs 0.70 ₽ / 4.20 ₽ per 1,000 (≈ $8.69 / $52.14 per 1M). In the descriptions of these models VseGPT says billing follows the provider token count, so we counted 1,000 price units as 1,000 tokens — counting by characters would make VseGPT even more expensive. Price list vsegpt.ru/Docs/Models on 01.11.2026, Bank of Russia rate 80.5573 ₽/$.
No subscription, no plans.
On VseGPT a subscription comes on top of the tokens: "Basic" is 199 ₽ per 30 days but excludes models marked "Professional" (Claude Sonnet / Opus, GPT-5.4 and others); "Professional" is 399 ₽ per 30 days or 99 ₽ per 3 days. Both have a default limit of 1 request per second (vsegpt.ru/Docs/Tariffs, 01.11.2026). RuAPI has no subscription fee: all models unlock after your first top-up, and only tokens are charged.
Claude Code and Codex — not in beta.
The Anthropic-compatible API of VseGPT is still in beta — its docs say it "may handle some requests incorrectly", and there are no plans to support v1/responses, which Codex CLI runs on (vsegpt.ru/Docs/API, 01.11.2026). On RuAPI the native Anthropic API is available at https://www.ruapi.ai for Claude Code, Cline and the Anthropic SDK, with Claude prompt caching; GPT models also have the Responses API, so Codex connects directly.
How to move from VseGPT to RuAPI in 3 steps
Bots, scripts and apps that talk to VseGPT through the OpenAI API need no rewrite — only the address, the key and the model IDs change.
Sign up for RuAPI
Sign up at www.ruapi.ai with email or in one click with Google / Yandex — no phone number needed. New accounts usually get $0.05 right away for testing: claude-haiku-5-5, deepseek-flash and gpt-6-luna work on it.
The other models unlock after your first top-up from 5 USDT — in USDT / USDC or with a card issued by a foreign bank, with no plan to choose.
Create an API key
In the console open API Keys and click "Create API key". It is handy to give each bot or script its own key with a spending limit, and the log shows the model, tokens and charge for every call.
Replace the address and model IDs
- OpenAI API: https://api.vsegpt.ru/v1 → https://www.ruapi.ai/v1.
- Claude Code: set ANTHROPIC_BASE_URL to https://www.ruapi.ai instead of https://services.vsegpt.ru:3000, and put your RuAPI key in ANTHROPIC_AUTH_TOKEN.
- Model IDs go without the provider prefix, and Claude versions use a hyphen: anthropic/claude-haiku-5.5 → claude-haiku-5-5, google/gemini-3.1-pro-preview → gemini-3.1-pro-preview. Drop the thinking-mode suffixes (-low, -medium, -high): anthropic/claude-sonnet-5.5-medium → claude-sonnet-5-5.
- For Claude Code, Cline, Kilo Code, OpenCode and Cherry Studio, the "Connect Claude Code, Cline & more" button in the console gives you a ready-made config with your key.
Your first request in a couple of minutes: docs.ruapi.ai/en/quickstart
Migration cheat sheet
API address
before: https://api.vsegpt.ru/v1
after: https://www.ruapi.ai/v1
Anthropic API: https://www.ruapi.ai (no /v1)
Model IDs
openai/gpt-5.4 → gpt-5.4
anthropic/claude-haiku-5.5 → claude-haiku-5-5
google/gemini-3.1-pro-preview → gemini-3.1-pro-preview
no openai/, anthropic/, google/ prefix
Payment
USDT — TRC20 / BEP20 / TON / Polygon
USDC — TRC20 / BEP20 / Polygon
Visa / Mastercard issued by foreign banks
from 5 USDT, no subscription
RuAPI prices for Claude, GPT, Gemini and DeepSeek
In US dollars per 1M tokens — pulled from the pricing page. VseGPT lists prices in rubles per 1,000: to compare, multiply by 1,000 and convert at the Bank of Russia rate.
| model | input $/M | output $/M |
|---|---|---|
| claude-sonnet-5-5 | $2.40 cache read $0.12 | $12.00 |
| claude-opus-5-5 | $4.80 cache read $0.24 | $24.00 |
| claude-haiku-5-5 | $0.12 cache read $0.0120 >100K: $0.60 | $0.60 >100K: $3.00 |
| claude-fable-5-2 | $9.60 cache read $0.24 | $48.00 |
| gpt-5.4 | $3.00 cache read $0.30 >272K: $6.00 | $18.00 >272K: $27.00 |
| gpt-5.6-terra | $2.40 cache read $0.24 >272K: $4.80 | $14.40 >272K: $21.60 |
| gemini-3.1-pro-preview | $2.40 cache read $0.24 >200K: $4.80 | $14.40 >200K: $21.60 |
| deepseek-flash | $0.17 cache read $0.0120 | $1.44 |
input — new input tokens, "cache read" — the part of the prompt read again from the cache, output — the model response. The cache discount applies to everyone, no subscription needed; for Claude it works in both the Anthropic and the OpenAI format (by cache_control marks; a client that sets none gets them from RuAPI when the next request of the conversation comes within 5 minutes).
The grey ">100K" row is the vendor price step for long prompts: if the prompt exceeds the threshold, the whole request is billed at that step.
RuAPI vs VseGPT, point by point
| RuAPI | VseGPT | |
|---|---|---|
| Claude Sonnet 5.5 (input / output) | $2.40 / $12.00 per 1M | 0.6 ₽ / 3.0 ₽ per 1,000 (≈ $7.45 / $37.24 per 1M) |
| Claude Opus 5.5 (input / output) | $4.80 / $24.00 per 1M | 1.0 ₽ / 5.8 ₽ per 1,000 (≈ $12.41 / $72.00 per 1M) |
| GPT-5.4 (input / output) | $3.00 / $18.00 per 1M (prompt up to 272K) | 0.70 ₽ / 4.20 ₽ per 1,000 (≈ $8.69 / $52.14 per 1M) |
| Gemini 3.1 Pro, prompt up to 200K (input / output) | $2.40 / $14.40 per 1M | 0.55 ₽ / 3.30 ₽ per 1,000 (≈ $6.83 / $40.96 per 1M) |
| DeepSeek Flash (input / output) | $0.168 / $1.44 per 1M | 0.05 ₽ / 0.15–0.20 ₽ per 1,000 (≈ $0.62 / $1.86–2.48 per 1M), "official" variants |
| Subscription fee | None — tokens only | 199 ₽ / 30 days ("Basic") or 399 ₽ / 30 days ("Professional") + tokens |
| Claude prompt caching | Yes: Sonnet 5.5 cache read — $0.12 per 1M | No cache-read price in the price list |
| Native Anthropic API (Claude Code) | Yes: https://www.ruapi.ai | Beta: https://services.vsegpt.ru:3000 |
| Responses API (Codex CLI) | Yes: GPT plus Claude, DeepSeek, Gemini and others | "No plans to support it" |
| Which model answers | Exactly the one named in the request | For new users "automatic optimisation" of requests to OpenAI models is on (can be switched off in settings) |
| Payment | USDT / USDC from 5 USDT or a card issued by a foreign bank (Visa / Mastercard) | In rubles: Russian bank cards, SBP, QR code; foreign cards from some countries, including the CIS; bank transfer for companies |
| Refunds | Unused balance — within 7 days after the top-up, on request to support | On request to support@vsegpt.ru: review takes up to 10 working days, money goes back to the same card |
| Works from Russia without VPN | Yes | Yes |
Sources (checked 01.11.2026): prices — vsegpt.ru/Docs/Models (Claude Sonnet 5.5 and Opus 5.5 — Thinking Medium variants, GPT-5.4 and Gemini 3.1 Pro Preview 200K — base variants, DeepSeek — "DeepSeek V4.1 Flash 1017 (official)"); subscriptions and limit — vsegpt.ru/Docs/Tariffs and vsegpt.ru/Docs/Tariffs/Limits; API — vsegpt.ru/Docs/API; payment — the vsegpt.ru home page; refunds — vsegpt.ru/docs/oferta, clause 3.7. VseGPT lists prices "per 1,000 characters", while its model descriptions say billing follows the provider token count; we counted 1,000 price units as 1,000 tokens — counting by characters would make the gap bigger. VseGPT also offers DeepSeek variants through alternative providers, including FP4-quantised ones, and some of them have cheaper output than ours. Bank of Russia rate 80.5573 ₽/$ on 01.11.2026; RuAPI prices are the base tier. Prices may change.