AIHow AI Works

Tag: Benchmarks

Claude Fable 5.1: Same $10/$50, Cache Reads Cut 75%

Anthropic shipped Claude Fable 5.1 and Mythos 5.1 on September 1, 2026. Sticker prices are unchanged; cache reads fall from $1 to $0.25 per million.

GPT-6 Astra Has No Realtime API and a 272K Pricing Cliff

Three days after launch, the GPT-6 Astra facts that decide a migration: no Realtime endpoint, a 272K billing threshold, and a quiet GPT-5.6 price cut.

Inworld Ships Realtime TTS-2 Voice Model Family

Inworld completes its Realtime TTS-2 rollout with a Flash variant and per-character pricing. Where the model actually ranks on the Artificial Analysis arena.

Meta Ships Muse Spark 1.3 With Big Long-Context Gains

Muse Spark 1.3 posts 98.5 on MRCR and cuts tokens 25% on coding — but cost per task rose 37%, and the headline scores use a mode that has not…

OpenAI Launches GPT-6 Astra With Record Benchmarks

OpenAI announced GPT-6 Astra on September 3, 2026, claiming 97.6% on FrontierMath Tier 4 v2 and 100% on ExploitBench. The ARC-AGI-3 result comes with a caveat.

Qwen’s E-Commerce Bench Runs AI Agents for a Simulated Year

Qwen released E-Commerce Bench: an LLM agent gets ¥100,000 and 365 simulated days to run online stores. 18 frontier models tested, none wins overall.

Best AI Image Model 2026: Price and Quality Have Decoupled

GPT Image 2 leads the arena, Google's cheapest model beats its priciest, and Midjourney is absent from the top twelve. What the data says in August 2026.

The Best AI Video Generator in 2026 Is Probably Not the One You Think

Sora is being switched off, Veo 3.1 ranks 11th, and Chinese models hold most of the top ten. What the video arena actually says in August 2026.

Qwen 3.7-Max: Alibaba’s Long-Horizon Agent Engine

Alibaba launches Qwen 3.7-Max, a flagship AI model demonstrating 35-hour autonomous operation, 10x kernel speedup, and cross-agent generalization.

Xiaomi MiMo-V2.5: The Next Generation of Open Agentic Models

Xiaomi announces the MiMo-V2.5 series, featuring flagship agentic performance and native omnimodality to rival frontier models like Claude 4.6 and GPT-5.4.