Person
Independent software developer and writer, co-creator of the Django web framework and Datasette; writes a widely read blog on AI/LLM developments.
Researcher cereblab's mitmproxy capture shows xAI's Grok Build CLI 0.2.93 sending about 5.10 GiB to storage and 192 KB to the model. The prompt said not to read files, so teams should audit what their agent CLIs transmit.
Reality
EcoHash reports that one vLLM flag took Qwen3.8-27B on a single RTX Pro 6000 from 999.9 to 1954.0 output tokens/sec. The figure comes from its own production config on vLLM 0.27.1, so it is one team's measurement on its own workloads.
Reality
Google reports 0.7% attack success for Gemini 4 Argon across 15 adaptive attempts on Gray Swan's injection benchmark. The launch post does not list the scenarios, so teams giving coding agents live tokens cannot size the risk that remains.
Reality
Anthropic's Claude Haiku 5.5 costs $0.10/$0.50 per million tokens up to 100,000 input tokens, then $0.50/$2.50 on the whole request. A dev.to breakdown advises sending Haiku only narrow, verifiable subagent work and keeping Sonnet 5.5 as the default coder.
Reality
Anthropic gave Max 5x subscribers $100 a month in API credits from 7 October, Max 20x $200 and Team plans up to $500 pooled. The credit excludes interactive Claude Code and expires each cycle, so the model a team runs sets what it buys.
Reality
OpenAI, Google and Vercel offer hard spend caps that reject requests, with Anthropic's in early access, a dev.to comparison found. The caps are monthly and enforced with a delay, and the post says none of them stops a single bad hour on its own.
Reality
Python 3.15 ends support for 3.10 and adds PEP 810 lazy imports, a keyword that code supporting 3.11 cannot contain. Simon Willison can now drop 3.10 and use 3.11 features, but the plugin startup fix needs a 3.15 minimum.
Reality
Anthropic launched Claude Haiku 5.5 at an average price about 75% below Haiku 4.5. The saving varies widely with prompt length, so teams moving classification, support or query traffic need to price their own requests before they switch models.
Perspective Coverage
22 publishersReality
Google's two new speech-to-speech models run tool calls in the background without breaking the dialogue, and Artificial Analysis prices them at $0.84 and $3.50 per hour of input audio on one benchmark subset.
Perspective Coverage
8 publishersReality
Google's new workhorse Flash is a model-string swap for anyone on AI Gateway, discounted through 31 December 2026. The claim worth testing is reduced tool-calling loop failures, not benchmark deltas.
Perspective Coverage
8 publishersReality
Alibaba's new 27B open-weight model defaults to its deepest reasoning setting. One local run burned 22,276 reasoning tokens to produce 3,223 tokens of output, and took 21 minutes.
Perspective Coverage
6 publishersReality
OpenAI's Decisions API answers yes/no, pick-one and scale questions for $0.10 per million input tokens, with output free. Jev sells the same three question types for text at less than half that, so OpenAI's case rests on image input and compliance terms.
Perspective Coverage
4 publishersReality
OpenAI disclosed the incident on September 16 under a framework it polices itself. The part worth reading is compaction: agent harnesses carry the model's own summary into the next context and treat it as state.
Perspective Coverage
12 publishersReality
Stack Overflow's 2026 survey of 30,903 developers found nearly half complacent in their jobs, against a third unhappy and just over 22% happy. Burnout led the reasons the complacent gave, so part of the largest group's problem is workload, which managers have some say over.
Reality
Vercel Labs' experimental ScriptC compiles TypeScript to native binaries that started in 1.78ms against Node's 61.78ms in one benchmark. The gain holds for short, statically typed programs, while framework code and sustained compute both ran slower than Bun or Node.
Reality
Cloudflare now sells web search from Ceramic.ai, Exa and Linkup through AI Gateway at partner list prices, billed from gateway credits with no markup. Teams already on the gateway skip a separate search vendor account, though their agents still have to issue each search call.
Reality
Diogo Almeida's TypeSafe put Jev into early access on 15 September with $40m from DCVC, and it sells a calibrated confidence number on every answer as the thing that makes automation possible, with the evaluations behind that claim built in-house.
Reality
AWS's project spending limits pause a project at its cap and permanently delete its data after 90 days paused without action. The cap bounds what a runaway agent experiment can bill, so someone has to own recovery and keep independent backups.
Reality
Agent runtimes should refuse unit 501 of a 500-unit budget before the call runs, a dev.to post argues. The new AWS and Google Cloud spend caps then become the backstop behind a tighter limit in the agent's own call path.
Reality
AWS added a per-project spend limit on September 16, 2026 that pauses service at the monthly cap, following Google Cloud's July launch of Spend Caps. A cap that checks each request stops at once, while one built on lagging billing data keeps charging until a function fires.
Reality
Build · September 22, 2026 · Confirmed14 publishers
Build · August 15, 2026 · Confirmed4 publishers
Build · August 24, 2026 · Confirmed2 publishers
Invest · August 30, 2026 · Confirmed2 publishers
Build · August 30, 2026 · Confirmed2 publishers
Build · September 10, 2026 · Confirmed3 publishers
Build · September 10, 2026 · Confirmed5 publishers
Build · September 11, 2026 · Confirmed7 publishers
Invest · September 22, 2026 · Confirmed16 publishers
Build · August 28, 2026 · Confirmed5 publishers
Build · August 23, 2026 · Confirmed2 publishers
Build · August 19, 2026 · Confirmed2 publishers