Was hat sich in der KI verändert?

New model available: Google: Nano Banana 2.1

google has a new model (google/gemini-nano-banana-2.1) on OpenRouter.

New model available: Z.ai: GLM 5.3 FlashX Legal

z-ai has a new model (z-ai/glm-5.3-flashx-legal) on OpenRouter.

New model available: Meta: Muse Spark Open

meta has a new model (meta/muse-spark-open) on OpenRouter.

New model available: Mistral: Mistral Large 4

mistralai has a new model (mistralai/mistral-large-4-0) on OpenRouter.

New model available: Anthropic: Claude Fable 5.2

anthropic has a new model (anthropic/claude-fable-5.2) on OpenRouter.

New model available: DeepSeek: DeepSeek V4.1 Flash 1017

deepseek has a new model (deepseek/deepseek-v4.1-flash-1017) on OpenRouter.

New model available: Google: Gemini 4 Carbon

google has a new model (google/gemini-4-carbon) on OpenRouter.

New model available: Meta: Muse Ember

meta has a new model (meta/muse-ember) on OpenRouter.

New model available: MoonshotAI: Kimi K3.1 Pro

moonshotai has a new model (moonshotai/kimi-k3.1-pro) on OpenRouter.

New model available: MiniMax: MiniMax M3.1 Flash

minimax has a new model (minimax/minimax-m3.1-flash) on OpenRouter.

New model available: Google: Gemini 4 Argon

google has a new model (google/gemini-4-argon) on OpenRouter.

New model available: Anthropic: Claude Haiku 5.5

anthropic has a new model (anthropic/claude-haiku-5.5) on OpenRouter.

New model available: OpenAI: GPT-6.1 Terra Pro

openai has a new model (openai/gpt-6.1-terra-pro) on OpenRouter.

New model available: OpenAI: GPT-6.1 Terra

openai has a new model (openai/gpt-6.1-terra) on OpenRouter.

New model available: Google: Gemini Omni 1.1 Flash

google has a new model (google/gemini-omni-1.1-flash) on OpenRouter.

New model available: Google: Gemini Omni 1.2 Flash

google has a new model (google/gemini-omni-1.2-flash) on OpenRouter.

New model available: Meta: Muse Wonder

meta has a new model (meta/muse-wonder) on OpenRouter.

Lindy price dropped: $49.99/mo → $29.99/mo

The cheapest paid plan for Lindy moved from $49.99/mo to $29.99/mo (-40%), verified on the vendor's pricing page on 2026-11-06.

Replit price dropped: $25/mo → $20/mo

The cheapest paid plan for Replit moved from $25/mo to $20/mo (-20%), verified on the vendor's pricing page on 2026-11-06.

Hermes Agent price rose: Free → $20/mo

The cheapest paid plan for Hermes Agent moved from Free to $20/mo, verified on the vendor's pricing page on 2026-11-06.

Suno price rose: $8/mo → $10/mo

The cheapest paid plan for Suno moved from $8/mo to $10/mo (+25%), verified on the vendor's pricing page on 2026-11-06.

Higgsfield price dropped: $19/mo → $9/mo

The cheapest paid plan for Higgsfield moved from $19/mo to $9/mo (-52.6%), verified on the vendor's pricing page on 2026-11-06.

Attio price rose: $44/mo → $50/mo

The cheapest paid plan for Attio moved from $44/mo to $50/mo (+13.6%), verified on the vendor's pricing page on 2026-11-06.

Cursor is now sold as part of the X Subscription, from $8/mo

Cursor's own pricing page now lists Free, Plus $8, Premium $30, Super $100 and Ultra $200 a month, plus Teams from $30 a user, sold through X Subscription Corp and shared with Grok. The old standalone Pro $20 and Pro+ $60 tiers are gone, and third-party models are billed separately by usage. Checked 2026-10-20.

Meta Muse Wonder 1.0 is in preview on the Meta Model API

Meta's developer docs list muse-wonder-1.0 as a preview model with a 1,048,576-token context window, available on the Model API and in Muse Code. We have not confirmed an API price on a Meta page.

Qwen-Image-2.1-Turbo: 7B image model in 8 steps, research license

Alibaba's Qwen-Image-2.1-Turbo is a 7B image generation and editing model that needs 8 denoising steps. Weights are on Hugging Face under the Qwen Research License Agreement, so check the terms before commercial use.

Grok Imagine Video 1.5 Lite costs $0.02 per second

xAI's model page lists grok-imagine-video-1.5-lite at $0.020 per second, against $0.080 for Grok Imagine Video 1.5 and $0.050 for the original Grok Imagine Video.

ChatGPT Pro gets parallel agents in GPT-6 Pro

A 16 Oct 2026 ChatGPT update lets GPT-6 Pro in the Chat tab run several agents on different parts of one task at once. It reaches eligible Pro users first; OpenAI published no speed or cost figures.

OpenAI previews GPT-6 Luna Decisions, a typed-answer API

OpenAI's changelog lists a beta Decisions API (v1/decisions, 6 Oct 2026) on gpt-6-luna that returns typed answers instead of free text and, per OpenAI, runs 10x faster than the Responses API. No price is listed yet.

Grok Voice 3.0 becomes the default voice model at $0.08 a minute

xAI's release notes say Grok Voice 3.0 (October 2026) now sits behind grok-voice-latest and is priced like Voice 2.0, which is deprecated. xAI's model page lists speech to speech at $0.08 per minute.

Meta releases Muse Spark Open, an 816B open-weight model

Meta published Muse Spark Open on 15 Oct 2026: 816B total and 42B active parameters, 1M-token context, under the Muse Model License with limited commercial terms. It is served by hosts such as Together AI, Fireworks AI and OpenRouter rather than Meta's own API.

DeepSeek refreshes deepseek-flash as V4.1 Flash 1017

The deepseek-flash id has served DeepSeek-V4.1-Flash-1017 since 17 Oct 2026. Calls need no change, DeepSeek says it is faster with shorter outputs, and the weights are open. No new price was announced.

MiniMax M3.1 Flash launches at $0.15 / $0.60 per 1M tokens

MiniMax released M3.1 Flash on 19 Oct 2026, priced at $0.15 input, $0.60 output and $0.015 cached input per 1M tokens. MiniMax says the open weights will come later. Benchmark numbers are MiniMax's own.

Nano Banana 2.1 image model is generally available

Google's gemini-nano-banana-2.1 (GA since 6 Oct 2026) improves text rendering and adds panoramic aspect ratios. Standard API price is $0.0336 per 1K image, $0.0504 per 2K and $0.113 per 4K; batch is half. The older gemini-3.1-flash-image is deprecated.

Gemini 4 Argon is now generally available, with a promo price until 30 Nov

Google made gemini-4-argon generally available in the Gemini API on 13 Oct 2026. It costs $2 per 1M input and $10 per 1M output tokens through 30 Nov 2026, then $4 and $20 from 1 Dec 2026.

MiniMax-Preis gestiegen: $20/mo → $22/mo

Der günstigste kostenpflichtige Tarif für MiniMax stieg von $20/mo auf $22/mo (+10 %), verifiziert am 25.09.2026 auf der Preisseite des Anbieters.

Neues Modell verfügbar: SpaceXAI: Grok 4.7

x-ai hat ein neues Modell (x-ai/grok-4.7) auf OpenRouter.

Grok 4.6: gleicher API-Preis $2/$6 wie 4.5, AA Index 61

xAI hat Grok 4.6 am 12. August 2026 zum Listenpreis von Grok 4.5 veröffentlicht. Von 4.5 mit einem Modellnamen-Tausch migrieren. Noch nicht auf unserem LiveBench-Snapshot.