AI.info
AI Tools Updates — Page 93
What changed in the AI tools this week — dated, from the makers’ own changelogs.
En anglais
- Tonic Textual — v493
In self-hosted deployments, the Textual Agent JavaScript sandbox tool is now optional.
- RunPod Pods — AstaBrief 8B on Serverless
AstaBrief 8B from Allen AI is available on Runpod Serverless with a validated vLLM recipe.
- CivitAI — Weekly Update: September 26 to October 2, 2026
FLUX 3 Image is in the generator. Text to image at 1K, 2K or 4K across 14 aspect ratios, and editing with up to 10 reference images.
- Botpress — Features & enhancements
You can now toggle whether a custom field is shown in the ticket details view while keeping it available for playbooks and tool calls.
- QA Wolf — v1.38.0
Adds commands to list failed screenshot comparisons, accept or restore baselines, opt out of investigations and remove issue reproductions.
- RunPod Serverless — New Release AstaBrief 8B on Serverless
AstaBrief 8B from Allen AI is available on Runpod Serverless with a vLLM recipe.
- OpenRouter — Model Router Benchmarks
OpenRouter now benchmarks model routers across six suites and provides a Router Index weighing quality, speed, and cost.
- Tonic Textual — In self-hosted deployments, the Textual Agent JavaScript sandbox tool is now optional.
- LocalAI — v4.11.0
Added audio scenes, ordered failover chains, /v1/systemone decision models, signed OCI galleries, Kimodo text-to-animation, and Studio and operations pages.
- Swapcard — Sherlock Goes Off-App: Embed Your AI Agent on Any Website 🕵️
Sherlock can be embedded on websites in guest mode, using only content marked available for guests.
- Higgsfield AI — Higgsfield API cashback
Higgsfield API added a $15 bonus after $100 in API spend and up to 20% cashback based on 30-day spend.
- Higgsfield AI — AI Influencer
AI Influencer lets users create AI characters and use them in videos with Motion transfer or Object swap.
- Readwise Reader — 🔊 NEW! Lifelike Voices (beta)
Reader's text-to-speech now offers 16 voices on web and in the mobile app, in American and British English.
- Readwise Reader — 🎧 Improved Text-to-Speech Controls
Moved playback controls into a floating player that shrinks into a pill as you scroll. The Return and Jump arrows point the right way again.
- Readwise Reader — 🧾 Fixed Upgrade Confirmation
Tapping Return or the X closes the screen and drops you back into your library.
- Readwise Reader — 📂 Fixed OPML Imports on Android
Taught the app to accept OPML files, so you can import an OPML file from your Files app on Android just like on iOS.
- Readwise Reader — 🎨 NEW! Reading Themes in Reader
Pick a reading theme from the Appearance sheet: Crisp, Sepia, Soft, Slate, Ink, Night, or PSL.
- Readwise Reader — 📱 Improved Android Navigation Bar
Extended the reading view and bottom sheets all the way down, so the page runs edge to edge.
- Readwise Reader — 🌓 Improved High Contrast Mode
Made secondary labels full black or white along with everything else.
- Readwise Reader — 🚀 Improved Reader Launch
Replaced all of that with a single launch screen and trimmed the work the app does on startup, so your library appears sooner.
- Taskade — v6.269.0: A New Model for TSK-Auto & Workspace Overview
TSK-Auto and TSK-Standard use a newer model, an older custom model is retired, and the model menu adds an Auto switch.
- Deepgram — Update Nova-3 Keyterms Mid-Stream
Send a keyterms array in a Configure message to replace the list during a /v1/listen session.
- Abnormal Inbound Email Security — This Week in Detection
Enhanced detection of lateral phishing from compromised senders, strengthening protection for internal-to-internal messages
- Abnormal Account Takeover Protection — This Week in Detection
Enhanced detection of lateral phishing from compromised senders, strengthening protection for internal-to-internal messages.
- Core42 — K2 Horizon Removed
K2 Horizon 375B is no longer available in Compass.
- Empathic Voice Interface (EVI) — TTS and EVI API sunset notice
Hume is sunsetting the TTS and EVI APIs. Access ends November 13, 2026 at 12:01 a.m. EST.
- Fern Docs — Draggable Ask AI chat widget
Visitors can move the standalone AskAiChat widget anywhere on the page; dragging is on by default in @fern-api/search-widget 0.6.0.
- Tavily Search API — How Insurance Teams Use AI Web Retrieval to Accelerate Underwriting and Risk Decisions
See how insurance teams use AI web retrieval to accelerate underwriting, enrich renewal profiles, and scale research.
- Replit — Workspace settings
Set rules company-wide, make exceptions for specific teams (Workspaces), and let team admins (Workspace admins) manage certain settings on their own.
- Replit — Jev is now available through Replit AI Integrations
Build apps that classify content, route requests, and score leads with fast, structured AI decisions, without managing API keys.
- Replit — New models from OpenAI and Anthropic
You can now choose GPT-6.1 Sol and Claude Sonnet 5.5 when building with Agent.
- Replit — New Settings page
Settings now opens as its own page instead of a modal. Integrations and Security get their own pages too.
- Truss — Baseten CLI 1.1.0
You can now rename a model from the terminal with baseten model rename. The model keeps its ID, endpoints, and deployments.
- Botpress — CSAT survey localization
CSAT survey questions, reminders, and rating labels now display in the ticket’s language for all supported Desk languages, with English fallback for unknown languages.
- Botpress — Ticket field interactions
Ticket detail fields now use consistent single-click-to-copy and double-click (or pencil) to edit behavior with unified styling across editors.
- Botpress — Ticket field visibility
You can now toggle whether a custom field is shown in the ticket details view while keeping it available for playbooks and tool calls.
- Botpress — Manual integration install
You can now install Desk bot integrations using a manually entered configuration in Settings Integration Hub.
- Grok — grok-voice-transcribe-1.0 end of life
All requests to that slug are routed to grok-voice-transcribe-2.0 at the same price.
- Legora — Your iManage, in Legora
Work with your iManage files directly in Legora, from finding the right document to saving your edits back.
- Fireworks AI — Higher serverless rate limits for Small models
We increased the adaptive rate-limit ceilings for Small serverless models and expanded the Small tier to cover models with less than 600B total parameters.