It was a busy week for frontier AI. Google shipped a new workhorse model and an ambitious open-source push to make AI practical on encrypted data; China’s DeepSeek rolled out peak/off-peak API pricing to match the explosion in demand; Anthropic published a detailed cost-efficiency playbook for Claude Code; and a widely-discussed essay asked whether AI’s edge on mathematics is less about smarter reasoning and more about an almost limitless working memory. Here are the top five AI stories of the day.

Google unveils Gemini 3.7 Flash, its “most intelligent workhorse model”

Just three weeks after Gemini 3.6 Flash, Google released Gemini 3.7 Flash (model ID gemini-3.7-flash), calling it the most intelligent model yet in its high-volume “Flash” line and framing it as a coding-and-agent model first, a chat model second. The update is a direct result of developer feedback and algorithmic optimizations, and Google has positioned it to compete on price as well as quality.

Benchmarks show a modest but steady improvement over 3.6 Flash, with “strong gains” in debugging and issue resolution, better design adherence in UI generation, and improved reasoning in knowledge-dense fields like finance, law, and bioscience. Behaviorally, Google says 3.7 Flash adapts better to roadblocks, asks clarifying questions when intent is ambiguous, and follows instructions with greater fidelity. It maintains the same 1M-token input context and 64k output limit, with multimodal input across text, image, video, audio, and PDF. Updated safeguards ship against misuse in CBRN (chemical, biological, radiological, nuclear) and cyber-offense domains.

Under the headline of the week is the price. Google is offering an introductory price of $0.75 per 1M input tokens and $3.75 per 1M output tokens through the end of 2026 — half the cost of the original 3.6 Flash pricing. The model is live in the Gemini API, Google AI Studio, Android Studio, Google Antigravity, the Gemini Enterprise Agent Platform, and consumer-facing via Spark. Customers including Box, Databricks, Harvey, LangChain, and Stanford’s Department of Biology offered testimonials. Why ship Flash before the long-awaited 3.5 Pro? Improving the model most production clients touch daily keeps Google in the release-cycle conversation during a burst of rapid shipping across every major lab.

Google open-sources HEIR to make private AI practical

Google’s Security Blog (by Staff Software Engineer Jeremy Kun) detailed how the company is making homomorphic encryption practical for AI. Homomorphic encryption allows computation directly on encrypted data, letting a cloud service process user inputs without ever decrypting them. Google is building the open-source HEIR (Homomorphic Encryption Intermediate Representation) compiler to convert pre-trained AI models that run on plaintext into versions that operate on ciphertexts.

HEIR is designed as a “one-click” solution so non-experts can add encrypted inference to production without needing a team of cryptographers. Google teamed with hardware accelerators Belfort, Niobium, Cornami, and Optalysys, and the project has become a research platform with collaborations at Georgia Tech, Carnegie Mellon, UC Santa Barbara, Purdue, Tsinghua, the University of Edinburgh, and others. The team released demonstrations including encrypted credit card fraud detection, network-threat detection, and a hotword detector. Critics on Hacker News noted fully homomorphic encryption has long been “horrifically slow” and that demos must be significantly stripped down, but Google argues the computational cost is falling rapidly — moving the privacy/security trade-off from “impossible” to “how much will you pay for it.”

DeepSeek introduces peak and off-peak pricing as V4 hits GA

Chinese AI leader DeepSeek updated its API pricing with a new peak/off-peak model, announced alongside the GA (general availability) release of its V4 lineup. Under the new structure, off-peak rates are 50% lower than peak rates, a pricing framework already familiar from cloud services but rare in the LLM API market.

Peak hours are 01:00–04:00 and 06:00–10:00 UTC. For deepseek-v4-flash, off-peak effective prices are $0.22 per 1M cache-miss input tokens and $0.66 per 1M output (half of the peak rates); for deepseek-v4-pro, the corresponding off-peak rates climb to $0.66 and $1.98. The expanded peak/off-peak model is aimed at shifting flexible workloads to cheaper off-peak windows, and mirrors a broader trend observers say is inevitable as inference demand scales and datacenter utilization becomes a key competitive lever.

Anthropic publishes an efficiency playbook for Claude Code sessions

Anthropic’s Lydia Hallie published “Maximizing the Value of Your Claude Code Sessions,” a practical field guide to getting the most out of every token spent on agentic coding. Why it is central to the AI conversation: with subscription and API pricing, a finished task has a price tag, and prompt engineering at the session level now directly shapes developer cost.

Key recommendations: run /clear between tasks to avoid til irrelevant context being re-sent; set your model and effort level (/model, /effort) before starting, since switching mid-conversation busts the prompt cache and forces a full re-prefill at full price; @-mention files instead of naming them to skip extra read calls; add quiet flags to noisy commands or run them in a subagent; and use /compact before a long break since the prompt cache expires after an hour on subscription (five minutes on an API key). Hallie explains the underlying token pricing mechanics (input vs output, prefill vs decode) and why one “square fix” can cost different amounts of tokens depending on how many files the model read along the way.

Is AI’s math edge really about working memory large humans can’t match?

A popular essay by Davide Piffer (titled “AI Isn’t Outthinking Mathematicians. It’s Out-Remembering Them.”) argued that the main AI advantage on math may not be superior reasoning but a virtually unlimited symbolic working memory — the giant context window and impressive external “notebook” that lets a model hold the whole problem, hundreds of intermediate equations, abandoned branches, and constraints all at once.

The post reviews psychological research on how a limited human working memory constrains mathematical performance — including Alloway and Alloway’s six-year longitudinal study in which early working-memory performance predicted later numeracy even after controlling for IQ, plus other studies showing working memory predicts mathematical achievement beyond general intelligence. Piffer’s argument, which drew more than 400 comments on Hacker News, is that part of what we label “machine intelligence” actually reflects the nearly unlimited symbolic workspace humans didn’t evolve with, much as scratch paper expands effective working memory by letting us externalize what we’re juggling. He concedes the context-window advantage is not equally useful across all forms of reasoning, and that advertised context length is not the same as perfectly usable memory — models can still lose track — but the sheer difference in potential capacity remains enormous.

Together these stories trace the shape of 2026: model vendors competing on cost and cadence, AI hardware and infrastructure, privacy-pressure on approaches like homomorphic encryption, riding agentic tooling, and a deeper, philosophical debate about what we actually mean when we say an AI is “smart.”

Another busy day in artificial intelligence, headlined by a fresh batch of model releases and infrastructure moves. Google introduced Gemini 3.7 Flash, its fastest-iterating “workhorse” series yet; OpenAI and Cerebras unveiled an “Ultrafast” inference tier for GPT-5.6 Sol; DeepSeek opened up its agent harness to developers; Mistral shipped OCR 4.1; and Google outlined a new open-source compiler designed to make privacy-preserving AI practical. Here are the five stories shaping the week in AI.

Google unveils Gemini 3.7 Flash, a faster and cheaper workhorse for coding and agents

Google has introduced Gemini 3.7 Flash, which the company calls its “most intelligent workhorse model yet” for coding and agents. The release lands just three weeks after Gemini 3.6 Flash and reflects direct developer feedback, according to Tulsee Doshi, senior director of product management at Google, writing on behalf of the Gemini team.

The model posts substantial gains in software engineering, knowledge work, and web development. On the FrontierCode 1.1 Main benchmark, it scores 43.6% versus 3.6 Flash 34.4%; on DeepSWE v1.1 it reaches 65.3% versus 49.0%. It also improved on knowledge-dense workloads — 34.0% versus 22.0% on the GDP.pdf document-processing evaluation, and 30.4% versus 17.0% on AutomationBench for real-world business workflows. In web development it reaches an Elo of 1588 on Arena.ai WebDev Arena, up from 1538.

Notably, Gemini 3.7 Flash launches at an introductory price of half the original 3.6 Flash cost per million tokens, and Google demoed it orchestrating sub-agents for tasks ranging from generating real-time 3D game assets to training robotics models via multimodal agent loops.

OpenAI and Cerebras unveil “Ultrafast” mode for GPT-5.6 Sol

Cerebras and OpenAI have shared an early look at Ultrafast Mode, a new service tier launching first in the OpenAI API and powered by Cerebras hardware that delivers up to 750 output tokens per second without any quality compromise. The tier has opened to a select group of customers, with access expanding over time.

According to Cerebras benchmarks, GPT-5.6 Sol on Ultrafast runs roughly 11x faster than Anthropic Fable 5 and about 5x faster than Opus 4.8 on Fast mode. Cerebras put the model head-to-head on Humanity Last Exam, a 2,500-question benchmark: Sol Ultrafast worked through all questions in 11 hours and 11 minutes, while Claude Fable 5 took 78 hours and 27 minutes at comparable accuracy, a nearly 7x speedup. On GDP-Val, a benchmark for economically valuable knowledge work, Ultrafast delivered a 5.6x end-to-end speedup with no quality degradation.

DeepSeek launches an open-source agent harness

DeepSeek has released DeepSeek Harness in developer preview, an open-source agent harness built on the Cordis plugin system. Its tagline — everything is a plugin — reflects a design where every capability (models, tools, skills, sessions, sandboxes, storage, loops, scheduling, and the UI) can be swapped or recomposed via configuration without modifying source code.

The project emphasizes traceability: everything the model sees is recorded in an append-only session log, including system prompts, reasoning, tool calls, and subagent scheduling, which can be inspected via a Trajectory view. It ships with multiple runtime modes — Standard, Code mode, Minimal, and Creator mode for building custom agent presets. Developers can get started with: npx @deepseek-ai/dsh web.

Mistral ships OCR 4.1 with paragraph-level structure

Mistral AI has released OCR 4.1, the latest version of its document-understanding service. The release adds native paragraph-level bounding-box extraction, structural block labels, and block-level confidence scores — valuable for parsing mixed documents like invoices, contracts, and research papers.

OCR 4.1 is available as a public preview, with pricing at $4 per 1,000 pages and $5 per 1,000 annotated pages. It is accessible through Mistral dedicated OCR API endpoints including structured annotations and batched processing.

Google open-sources HEIR compiler to make private AI practical

Google has added HEIR (Homomorphic Encryption Intermediate Representation) to its Private Computing Toolkit — an open-source compiler designed to make cryptographically-secure private AI inference more practical. With homomorphic encryption, computations run directly on encrypted data, so a cloud service could deliver recommendations or analytics without ever seeing the underlying data.

Staff Software Engineer Jeremy Kun described the technology as rapidly maturing and shifting the privacy/capability trade-off from a security question to a question of cost — a cost that is rapidly decreasing. HEIR automates much of the work that previously required a team of cryptographers.

The bottom line

This week news underscores two converging themes across AI. The first is speed: OpenAI and Cerebras are racing on inference throughput, while Google continues its rapid Flash-series cadence pairing better benchmark scores with falling prices. The second is trust and control: DeepSeek pushes open traceable agent harnesses, Google invests in cryptographic privacy, and builders get cheaper faster ways to deploy frontier models.