Skip to content
Stork.AIstork.ai
New models
  • GoogleGemini 4 Hydrogen$1.50 / $7.50
  • GoogleGemini 4 Carbon$2 / $10
  • MetaMuse Ember$0.10 / $0.50
  • AnthropicClaude Fable 5.2$8 / $40
  • OpenAIGPT-6.1 Terra$1 / $5
  • MorphKimi K3.1 Pro$2.50 / $14
  • Moonshot AIKimi K3.1 Pro$2 / $10
  • Moonshot AIKimi K3.1 Pro Highspeed$4 / $20
  • MiniMaxMiniMax-M3.1-Flash$0.15 / $0.60
  • MistralDeepSeek V4.1 Flash 1017$0.30 / $1.20
  • MorphDeepSeek V4.1 Flash 1017$0.12 / $0.80
  • DeepSeekDeepSeek V4.1 Flash 1017$0.30 / $1.20
  • DeepSeekDeepSeek V4 Flash Vision Exp$0.15 / $0.60
  • DeepSeekDeepSeek V4 Flash$0.15 / $0.60
  • InceptionMercury 3$0.10 / $0.35
  • GoogleGemini 4 Argon$2 / $10
  • MetaMuse Wonder$4 / $20
  • AnthropicClaude Haiku 5.5$0.10 / $0.50
  • MistralMistral Large 4.0.8$0.70 / $2.10
  • MistralMistral Large 4$0.70 / $2.10
See all prices

For AI tool makers

Is ChatGPT recommending your AI tool?

Start with the free check: can ChatGPT, Claude and Perplexity read your site? Then Stork Pro tracks what they answer in your category, every week.

Free, no signup. · Looking for an AI tool instead? Search the directory ↓

Stork AI Daily

The lead

Google's TPU v7 delivers 1.85x more tokens per dollar than NVIDIA.

Wren Calloway·The Daily

NVIDIA's hardware moat is evaporating. Google is forcing the incumbent to rely on software tricks just to stay competitive.

Google's TPU v7 is officially outperforming NVIDIA's GB300 by 1.85 times on tokens per dollar in an FP8 comparison. This is a direct hit to the core of NVIDIA's pricing power, striking right at the metric that matters most to production AI: inference cost. For years, the market accepted that NVIDIA's premium pricing was justified by unmatched performance. Google has just shattered that consensus by delivering nearly twice the token efficiency per dollar.

Read the full edition →

TL;DR

  • Google's TPU v7 delivers 1.85 times more tokens per dollar than NVIDIA's GB300 in FP8.
  • Anthropic grew revenue 13 times while compute expenses only grew 4.3 times.
  • OpenAI is building custom Jalapeno chips to break free of the GPU monopoly.
  • United Airlines banned third-party AI agents from booking flights to stop middlemen.
  • Elon Musk's xAI dumped JAX for a custom C++ stack to train Grok 4.8.
  • We compared ten AI watermark removers on price, limits, and the law.

From the blog

The rest of the field

From the editor

The GPU monopoly is officially bleeding, and Google just delivered the deepest cut yet. If you thought NVIDIA's dominance was an immovable law of physics, it's time to check the math. For the last three years, the entire artificial intelligence industry has operated under a single, depressing assumption: you pay whatever Jensen Huang asks, or you do not get to train frontier models. That era is over.

Read the full edition →

Previous edition: Anthropic targets $1.404 trillion IPO valuation

AI tools directory

Find the right AI tool

Search the catalog, browse by category, or start with the tools we have reviewed in depth: real prices from vendors' own pages, alternatives and user opinions.

Browse by category

Newly reviewed

Prices checked on the vendor's own pricing page

Just landed

  • VoiceConnect — VoiceConnect is an AI Business Operations Platform that manages calls, outreach, and records. It answers…
  • CompreFace — Self-hosted open-source face recognition service with a REST API and web UI for face detection, verification…
  • InsightFace — Open-source 2D and 3D face analysis toolkit (ArcFace, RetinaFace); code is MIT, but its pretrained models are…
  • DeepFace — Open-source Python library for face verification, recognition and attribute analysis that wraps models such…
  • Luxand.cloud — Face recognition API for developers: enroll people, then verify or identify faces in photos and video, with…
  • Megvii Face++ — Face++ cloud API from Megvii for face detection, comparison, search and facial attributes, with SDKs for…

New in the directory — list your tool today and it appears here.

Browse the full directory

From the blog →

Text-to-Speech APIs for Phone Calls: 19 Voices Tested on Latency, Numbers and Price (Nov 2026)

Best Call Center Software (2026): 20 Platforms by Team Size, With Real Prices

About Stork

Stork.AI is an AI tools directory with a daily front page of AI news. Tool pages state real prices from each vendor's own pricing page, alternatives and user opinions; a free check reads any site the way AI crawlers do, and Stork Pro tracks how ChatGPT, Claude, Perplexity, Gemini, Grok and Google answer in your category. Stork also lists MCP servers and vetted AI agencies, and publishes in eight languages.

Editions: English·Русский·Español·Français·Deutsch·日本語·한국어·Português