Bookmark this page: press Ctrl+D to keep it handy — new model launches and promotions are announced here first.
🔥 Latest Updates
New Model · Google
Nano Banana 2.1 Is Live
$0.05 per Request, 4K Priced Like 1KGoogle’s October 6 upgrade to Nano Banana 2,gemini-nano-banana-2.1, brings better image quality, text rendering and multi-turn consistency. Token-based billing is $0.66 input / $13.2 output per 1M tokens, and 512px is not supported; Nano Banana 2 stays available at the same price.📖 Read details | 📚 Read the docs | ⚡ Live updateNew Model · Microsoft
Microsoft MAI-Image 2.6 Is Live
$0.12 / $0.06 per Image, Any SizeMicrosoft AI’s in-house image modelsMAI-Image-2.6 and MAI-Image-2.6-Flash are live via the official channel, ranked No. 2 on Arena for both text-to-image and editing, with strong Chinese text rendering. Text-to-image and reference-image editing with custom width + height, available in the default group.📖 Read details | 🖼️ Integration docsNew Model · AZ8
Oxygen Video Generation Is Live
$0.02 per Second at Any ResolutionAPIYI and AZ8 launch AZ8’s in-house all-round video modeloxygen-1.0: an OpenAI Videos-compatible API with one model for text, first-frame, first-and-last-frame, and reference image/video/audio video, 320p–768p, 4–15 s. Priced at $0.02 per second (a 5-second clip is $0.10), available in the default group and built for low-cost volume.📖 Read details | 🎬 Integration docsNew Model · OpenAI
GPT-6.1 Sol Is Live
Provider Pricing, Cache Reads Half of GPT-6 SolOpenAI’s upgraded Sol, launched at DevDay on 29 September, scores 52 on the Artificial Analysis Intelligence Index, just 1 point below GPT-6 Astra. Priced at $2 input / $10 output in line with the provider, with cache reads cut from $0.20 to $0.10; savings come through our top-up promotions.📖 Read details | 🔗 PromotionsNew Model · Anthropic
Claude Sonnet 5.5 Is Live
Opus 5.5-Level Coding at Sonnet 5 PricingReleased by Anthropic on 28 September and more than 30% faster than Sonnet 5. It scores 70.6% on Terminal-Bench 4.0 (Anthropic), above Opus 5.5. Priced at $2/$10 per million tokens, matching the official price; replacethinking: disabled with between_tools.📖 Read details | 🔗 PromotionsNew Model · MiniMax
MiniMax H3 Video Generation Is Live
$0.03 per Second, Free Reference MediaMiniMax’s omni-modal video model, released 31 July, now runs on APIYI as a self-hosted deployment of the open weights: one endpoint for text, first/last-frame, and mixed image/video/audio references, with stereo sound, 768P, 4–15 s. Priced at $0.03 per second (the official MiniMax API charges $0.08), available in thedefault group.📖 Read details | 🎬 Integration docsNew Model · ByteDance
Seedream 5.0 Flash Is Live
$0.018 per Image, the Cheapest in the SeriesBytePlus’s new fast image modelseedream-5-0-flash-260915 focuses on speed, layout and low cost. It generates in 15–20 seconds in our tests, with every line of a Chinese poster rendered correctly. Priced at $0.018 per request, matching the provider, with free reference images; top-up bonuses bring it roughly 9%–17% below the official price.📖 Read details | 🔗 PromotionsNew Model · xAI
Grok 4.7 Is Live
A New 2.1T Base Model at the Same PriceSpaceXAI’s new flagship, released 21 September, grows from 1.5T to 2.1T parameters, with the biggest gains in coding agents: Terminal-Bench 4.0 rises from 18% to 33%. Priced at $2/$6 per 1M tokens, matching the provider and the previous generation, though it uses about twice the output tokens per task as 4.6, so estimate cost on real tasks.📖 Read details | 🔗 PromotionsNew Model · OpenAI
GPT-6 Sol and Luna Are Live
Half the Previous Price, PermanentlyOpenAI’s GPT-6 mid and light tiers, released 22 September, the same day Claude Opus 5.5 cut prices.gpt-6-sol is priced at $2/$10 and gpt-6-luna at $0.10/$0.50 per 1M tokens, about half of the 5.6 generation, with every line item matching the provider. Both Chat and Responses accept function tools in our tests, so migrating from 5.6 is a model-name change.📖 Read details | 🔗 PromotionsNew Model · Anthropic
Claude Opus 5.5 Is Live
Fable-Class Results for Less Than Opus 5Anthropic’s next-gen Opus, released 22 September, scores 66.4% on Terminal-Bench 4.0, ahead of Fable 5.1’s 55.8%. Priced at $4/$20 per 1M tokens with cache reads cut to $0.20; every APIYI line item matches the provider. Note: thinking can’t be disabled, forced tool use returns 400, and the default effort is nowmedium.📖 Read details | 🔗 Promotions📖 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