On 13 November 2026 Google said monthly Developer Program Cloud credits can be claimed inside Google AI Studio. Only the primary Google AI Pro or Ultra subscriber can use them, and Gemini API Cloud billing requires a US$5 minimum prepayment and balance. Monthly amounts by plan are not listed.
On 13 November 2026 Microsoft published a plan to move the on-device language model from Phi Silica to Aion Instruct. Official figures: Aion Instruct 2.3B parameters and a 32K-token input limit; Phi Silica 3.8B and 3.5K tokens. Insider Preview is planned from 20 November 2026; retail is targeted for late January 2027.
On 13 November 2026 OpenAI announced ChatGPT Wallet, which lets dots complete a purchase after the user approves it. The first release is limited to US ChatGPT Pro subscribers aged 18 or older with an eligible US-issued card. Approval alone does not confirm a merchant order.
Five original text exercises: coupon order, cross-zone dates, caption reading rates, image alt text and quotations versus paraphrases. Fictional fixtures and human checks, not tested results. Includes official Gemini web eligibility and limits for Hong Kong, Macau and Taiwan.
Supply an offline game, starting state and budget to design tests that separately record reasoning, input delay and human intervention.
Supply a reproducible bug, allowed files and existing rules to propose minimal changes and tests, including old-save compatibility.
Use a fictional reminder, time and recipient to define evidence for draft, queued, sent, delivered and cancelled states in a test plan.
Use authorized recipe summaries to organize equipment, time and preference evidence without filling missing facts.
Give a text AI redacted configuration copies and allowed scope to draft a configuration-change and acceptance checklist for human review. This is a drafting exercise; the source used Codex.
Text exercise: Provide synthetic phrases, fictional entity states and allowed actions. Draft intent checks, confirmation points and refusal conditions.
Give a text AI an animation-state diagram and engine version to draft a transition, interruption and recovery test table for human review. This is a drafting exercise; the source used Codex.
Give a text AI synthetic notes and an approved sharing list to draft a shareable draft linked to approved evidence for human review. This is a drafting exercise; the source used Cursor.
Align dates, room types and quote scope without silently relaxing requirements when no option matches.
Use fictional booking screens and stay requirements to draft date and room checkpoints after navigation, without treating form entry as a booking.
Provide fictional meeting goals, area, time window, party size, budget and transport. Draft clarification questions and restaurant comparison fields.
Give a text AI authorized annotations, page references and bibliography to draft a research-note draft traceable to annotations for human review. This is a drafting exercise; the source used Claude Code.
Supply a fictional historical research log and source IDs to distinguish findings, failed searches, unsearched areas and conflicting identities.
Give a text AI authorized text, learner level and learning goals to draft a lesson specification and answer-check list for human review. This is a drafting exercise; the source used Claude Code.
Text exercise: Provide a test RSS feed for your own articles, processed IDs, audience and platform limits. Draft faithful promotional copy with duplicate-handling reasons.
Supply an authorized activity export and time range to draft a recap that distinguishes proposals, reviews, merges and releases with evidence.
Give a text AI synthetic error logs and redacted investigation notes to draft an internal escalation and customer-facing draft for human review. This is a drafting exercise; the source used Cursor.
Give a text AI query metrics, page content and existing titles to draft content-faithful title proposals and an evaluation plan for human review. This is a drafting exercise; the source used Claude Code.
Define dimensions and dependencies, then adjust and export a model in CADAM, using OpenSCAD for section checks where needed.
Using Claude where available: Supply test-model parameters, dependencies and saving steps to draft checks for changes and reopening.
Provide authorized test images, filenames and naming rules. Use an image-capable chat to suggest names and categories with visible evidence.
Text exercise: Provide a synthetic attendee list, supplied public professional profiles and a meeting purpose. Draft relevant briefs with conflicts and unknowns.
Give a text AI synthetic email threads and a reply-needed definition to draft a decision table of awaiting reply, handled and uncertain threads for human review. This is a drafting exercise; the source used Devin.
Supply fictional email tasks and available capabilities to draft a permission table separating reading, drafting, sending and deletion.