Skip to main content
Welcome to the APIYI changelog. This is where we announce new model launches, pricing changes, and feature updates. We’re committed to giving you the most capable AI services at the best value.
Bookmark this page: press Ctrl+D to keep it handy — new model launches and promotions are announced here first.

​
🔥 Latest Updates

​
New Model · OpenAI

​
GPT-6.1 Terra Is Live

Visual Reasoning and Writing at $1 Input / $5 Outputgpt-6.1-terra passed functional checks for text, streaming, JSON, images, and a Responses tool round trip. Standard input/output rates are $1/$5 per million tokens. See the launch guide for endpoint-specific cache observations and the tested input lengths.📖 Launch guide and API examples
​
Service Notice

​
Oct 26 Outage Review

Apology, Follow-Up Arrangements, and IP Allowlist ReminderThe October 26 incident review is now available, and we apologize to affected customers. A 10% discount will apply automatically to Default and SVIP usage throughout Wednesday, October 28 (UTC+8), with no application required. Customers who temporarily removed IP allowlists because of 401 errors should restore their original configurations.📖 Full incident review and compensation details
​
Docs Update · ByteDance

​
Seedance 2.5 Docs

Preview First, Then Generate the Final VideoThe Seedance 2.5 guide now covers a 480p draft with draft: true, followed by a 1080p final video referencing the draft ID. Both stages are billed separately, and the draft ID is valid for seven days.📖 Draft mode guide | 📘 Model overview | ⚡ Live update
​
New Model · Moonshot

​
Kimi K3.1 Pro Is Live

Standard and Highspeed Official Transfer, About 83.3% Effective Cost with BonusPro and Highspeed official-transfer models passed basic reasoning, streaming, tool-call and vision tests. Input/output rates match official pricing at $2/$10 and $4/$20 per million tokens. Eligible deposits with a 20% bonus reduce effective cost to about 83.33% of list pricing.📖 Integration guide | 🔗 Deposit promotions | ⚡ Live update
​
New Model · DeepSeek

​
DeepSeek Search and Fetch APIs Are Live

Billed per Success, Official Price with BonusNew deepseek-search and deepseek-fetch (/v1/tools/search, /v1/tools/fetch) give agents and RAG live web data without consuming tokens. Priced at $0.0009 per request; with the 20% top-up bonus large-deposit customers pay $0.75 per 1,000, the same as official, tax included. Failed requests are free; available in default / svip.📖 Read the docs | ⚡ Live update
​
New Model · xAI

​
Grok Imagine Official Is Live

Images and Video at xAI’s Own PricesA new GrokOfficial group (by application): images on grok-imagine-image / -quality / -2.0 are billed by resolution and quality tier at $0.02–0.08 each; video on grok-imagine-video / -1.5 / -1.5-lite covers text-to-video and image-to-video at $0.02–0.25 per second. The former Grok_imagine group is now Grok_imagine_Reverse (reverse channel), with unchanged pricing.🖼️ Official images | 🎬 Video generation
​
New Model · Google

​
Gemini 4 Argon Is Live

Google’s New Flagship, Same Accuracy With Far Less ThinkingThe first Gemini 4 flagship, on APIYI from day one. In our tests it matched 3.1 Pro at 36/36 with about 1/12 of the thinking tokens. Priced at the limited-time promotional rate of $2 / $10, moving to $4 / $20 with Google on 1 December; remove thinkingBudget when migrating.📖 Read details | 📘 Docs
​
Price Update · Google

​
Nano Banana 2.1 Per-Request Price Cut

$0.05 Down to $0.045, Same for 1K–4Kgemini-nano-banana-2.1 per-request billing drops 10%, from $0.05 to $0.045, the same price for 1K / 2K / 4K — for 4K, nearly 30% cheaper than token-based. Token-based pricing is unchanged, and top-up bonuses lower the cost further.📖 Read details | 🔗 Model docs
​
Service Notice

​
Personal Access Tokens Are Live

Legacy system tokens are scheduled to retire on 2026-11-11 (UTC+8)Balance, log and token management APIs now use personal access tokens, which support per-scope permissions, expiry dates and IP allowlists. Legacy system tokens are scheduled to retire on 2026-11-11 (UTC+8). We may extend the date based on usage, but recommend switching now: just replace the header value.📖 Read details | ❓ Personal access token guide
​
New Model · Alibaba

​
Wan3.0 Video Generation Launches

One model ID for text, image, reference and video-edit modeswan3.0-video (Standard) and wan3.0-video-prime (Prime) are live, with clips up to 30 seconds, 480P–1080P and audio by default. Priced at 98% of official; Standard at 720P costs $0.0588/sec, about 30% less than Wan2.7.📖 Read details | 🔗 Model docs
📖 These are the 10 most recent announcements. For everything earlier, see the announcement archive, browsable by month, category, or vendor.

Deep dives

Browse the AI Radar section

Live updates

Model status and service notices

Promotions

See current deposit bonuses