🚀 New
Claude_Enterprise group (identifier ClaudeOfficial) at a 1.2x rate — Anthropic official channel only (no AWS Claude), high cache hit rate, 12 Fable / Opus / Sonnet / Haiku models; priced on the same basis as image2Enterprise. Select this group in your token settings to use it.📖 View details🔐 Still told you can’t enroll 2FA or a passkey during the grace period? Sign out, sign in again, wait about one minute, then try again — the grace period is applied per sign-in session, and some older sessions are not released automatically; a fresh sign-in is released within about a minute, after which you can enroll from the account settings page.📖 View details
🚀 Nano Banana 2.1 (
gemini-nano-banana-2.1) is live at $0.05 per request, the same price for 1K / 2K / 4K — Google’s latest GA image model, with better quality, text rendering and multi-turn consistency than Nano Banana 2. Token-based billing is $0.66 input / $13.2 output per 1M tokens; 512px is not supported. Nano Banana 2 remains available at the same price.📖 View details🧾 Need last month’s bill exported? Ask in your dedicated WeCom group and we’ll deliver it after the 8th of each month — or submit a ticket on the site, and we’ll take good care of you.📖 View details
🔐 From now until Oct 15, 23:59 (UTC+8), you can enroll 2FA or a passkey right after signing in —— no 72-hour stable-environment wait. We suggest adding a passkey first; if you enroll 2FA first, sign out and sign in again with your password plus a 2FA code before adding a passkey, and save your backup codes.📖 View details
⚠️
gpt-image-2.5-sunburst-vip and gpt-image-2.5-flare-vip have been unstable lately due to occasional capacity shortages —— The official-reverse route depends on provider-side risk controls and account capacity; we keep maintaining it, and recommend client-side retries or rotating across models. gpt-image-2-vip, the official gpt-image-2.5-sunburst, or the 1K-only gpt-image-2.5-all are also available as needed.📖 View details📚 New FAQ: image calls behind a VPN drop in batches after 90-180 seconds and are still billed because the local proxy closes idle connections —— No data flows while an image is generated, so the proxy closes the connection on its idle timeout, and a longer client timeout does not help. Route
api.apiyi.com directly; direct server deployments usually never see this.📖 View details🚀 Microsoft
MAI-Image-2.6 / MAI-Image-2.6-Flash are live at $0.12 / $0.06 per image, any size —— Microsoft AI’s in-house image models, No. 2 on Arena for both text-to-image and editing, served via the official channel; text-to-image and reference-image editing, strong Chinese text rendering, custom width + height up to 1536×1536, available in the default group.📖 View details🚀
oxygen-1.0 video generation is live, priced at $0.02 per second at any resolution —— AZ8’s in-house all-round video model, launched with APIYI: OpenAI Videos-compatible, one model for text, first-frame, first-and-last-frame, and reference image/video/audio video, 320p–768p, 4–15 s, built for low-cost volume; the default group works.📖 View details💡 Calling
glm-5.3-flash or the Seed 2.0 series via the Responses API? Set your token group to CodexResponses —— on the default group some /v1/responses requests first land on a Chat Completions-only channel and are retried internally, adding seconds to tens of seconds; the dedicated group goes straight to a Responses channel. Chat Completions calls are unaffected.📖 View details📖 For earlier updates, visit the Live Updates Archive — browse history by month, category, or vendor.
Deep Dive
AI Radar
Telegram
Global users
Media Models
New additions