According to OpenRouter, open-weight models reached 60 percent of token volume in U.S.-originating requests and 65 percent in EU-originating requests by August 2026.

Overview Debates Related claims Topics Sources

Anthropic priced Claude Fable 5.2 and Claude Mythos 5.2 at $8 per million uncached input tokens and $40 per million output tokens, representing a 20% cost reduction over Fable 5.1 and Mythos 5.1, their predecessors

Anthropic priced Claude Fable 5.2 and Mythos 5.2 at $8 per million input tokens and $40 per million output tokens, representing a 20% cost reduction from generation 5.1

Apple is considering lowering daily AI token spending limits for non-engineering teams from $100 to as low as $25

Apple is planning further internal budget reductions that could decrease daily AI allowances for certain engineering teams to $100

In October 2026, Apple reduced the daily artificial intelligence spending allowance for some engineering employees from $300 to $200

Some Apple employees expressed concern that internal AI token budget cuts signal broader corporate cost reductions and potential workforce layoffs

Apple established a procedure allowing employees who require higher daily AI token allowances for work to request senior leadership exemptions

Microsoft restricted internal AI spending allowances, while Meta Platforms introduced token budgets to encourage employee adoption of in-house AI tools

Artificial Analysis reported that Inkling 2 Small generated 154 million output tokens during its Intelligence Index evaluation compared to a median of 89.8 million tokens among comparable models

Training on Thinking Machines' Tinker platform costs $1.50 per million tokens at a 64K context window or $3 per million tokens at a 256K context window

Tinker's serverless pricing for Inkling 2 Small is set at $0.30 per million input tokens, $0.03 per million cached input tokens, and $1.00 per million output tokens

Baidu released the FP8 and BF16 weights for ERNIE 5.5 Flash Base on Hugging Face and ModelScope under ERNIE 5.5 Flash Base License v1.0

Enterprise leaders and public company shareholders face increasing pressure to demonstrate measurable financial value and return on investment from rising token consumption and API expenditures

MIT CSAIL researchers introduced 'weight-space programming,' a technique enabling LLMs to interpret and directly edit neural network weights without conventional gradient-based training procedures

Cohere co-founder Ivan Zhang advised enterprises to treat open-weight AI models with the same security precautions as any unverified program downloaded from the internet.

Cohere called on governments to clarify whether their policy restrictions on Chinese AI models apply to open-weight models running on government-owned hardware.

Vercel's October 2026 Production Index reported that Chinese AI developers DeepSeek, Z.ai, and Moonshot AI processed a combined 68.8 percent of tokens on the Vercel AI Gateway during September 2026.

Google said Gemini 4 Carbon carries introductory pricing of $2 per million input tokens and $10 per million output tokens, which will later increase to $4 and $20 per million tokens respectively

Google offers a 95% discount on cached input tokens for Gemini 4 Carbon

Entire claims that using Continuous Memory and Entire Graph reduces AI agent token consumption by up to 71% by preventing agents from repeating resolved mistakes