The New Stack May 26
The context window has been shattered: Subquadratic debuts a 12-million-token window
thenewstack.io/subquadratic-12-million-context-window/
Story Details
| Last Updated | |
| Total Articles | 3 |
Main Topics
behavioral and cognitive process of selectively concentrating on a discrete aspect of information, whether deemed subjective or objective, while ignoring other perceivable information
Similar to This
Previous
Computerworld Oct 29
It costs a lot more to use AI models in some languages: The difference in token usage can be 17 times greater for certain languages, driving up costs and ...
computerworld.com/article/4241844/ai-models-are-much-more-expensive-for-less-common-languages.html
The New Stack Oct 17
“Take work away from frontier LLMs”: Subquadratic’s decision model claims 12x Jev’s context window
thenewstack.io/subquadratic-snap-jev-decision-model/
Your local LLM is quietly wasting RAM on context you never use, and one setting gives it back
xda-developers.com/local-llm-wasting-ram-context-never-use-one-setting-gives-back/
The New Stack Sep 23
OpenAI cut GPT-6 token prices in half. The bigger lever may be the cache.: OpenAI cut GPT-6 Sol and Luna API prices in half. The quieter savings come from ...
thenewstack.io/openai-prompt-caching-costs/
Token-maxxing is dead. Agentic memory is what comes next.: The context window is the scarce resource. The challenge isn’t stuffing more information into every ...
venturebeat.com/data/token-maxxing-is-dead-agentic-memory-is-what-comes-next
The New Stack Jul 08
“Nature is the most computationally efficient system we know”: How Refiant used swarm optimization to build a 10-million-token AI model
thenewstack.io/refiant-protea-swarm-optimization/
I made the mistake of loading every MCP server into my coding agent, but here's what I do now instead
xda-developers.com/made-mistake-of-loading-every-mcp-server-into-coding-agent/
The New Stack Jul 02
What comes after attention? This startup says it already knows.: CTO Alex Whedon says Subquadratic isn't a sparse-attention company at all ...
thenewstack.io/subquadratic-sparse-attention-no-attention/
The New Stack Jun 29
How we cut AI costs by 80%: Cut enterprise AI agent costs by 80% using a context lake. Learn how pre-joined data simplifies context windows and optimizes tokens.
thenewstack.io/how-cut-ai-costs/
A startup claims it broke through a bottleneck that’s holding back LLMs: Subquadratic has now shared more details about its new model. But some are still ...
technologyreview.com/2026/06/19/1139313/a-startup-claims-it-broke-through-a-bottleneck-thats-holding-back-llms/
Users say Gemini starts forgetting long before it’s supposed to: Google says AI Pro and Ultra users have a one-million-token context window ...
androidauthority.com/gemini-start-forgetting-context-window-3674304/
OpenAI voice models get GPT-5-class reasoning : OpenAI's Realtime-2, Realtime-Translate, and Realtime-Whisper split voice into discrete models ...
venturebeat.com/orchestration/openai-brings-gpt-5-class-reasoning-to-real-time-voice-and-it-changes-what-voice-agents-can-actually-orchestrate
Miami startup Subquadratic claims 1,000x AI efficiency gain with SubQ model; researchers demand independent proof.
venturebeat.com/technology/miami-startup-subquadratic-claims-1-000x-ai-efficiency-gain-with-subq-model-researchers-demand-independent-proof
github.com Mar 27
GitHub - Boof-Pack/token-enhancer: A local proxy that strips web pages down to clean text before they enter your AI agent's context window. 704K tokens → 2 ...
github.com/Boof-Pack/token-enhancer
The New Stack Mar 16
Anthropic makes a pricing change that matters for Claude’s longest prompts: Anthropic removes long-context pricing surcharge for Claude Opus 4.6 and Sonnet 4.6 ...
thenewstack.io/claude-million-token-pricing/
8 local LLM settings most people never touch that fixed my worst AI problems: If you run LLMs locally, these are the settings you need to be aware of.
xda-developers.com/local-llm-settings-most-people-never-touch/
OpenAI's GPT-5.4 sets new records on professional benchmarks: The new model introduces native computer use, a 1-million-token context window ...
thenextweb.com/news/openai-gpt-54-launch-computer-use-benchmarks
Anthropic offering AI model to all three government branches for $1: Anthropic announced Tuesday that it will offer its artificial intelligence (AI) model Claude ...
trib.al/5T1OYSD
Next