AI.info
AI Tools Updates — Page 166
What changed in the AI tools this week — dated, from the makers’ own changelogs.
Auf Englisch
- Lokalise — AI scoring in workflows is now opt-in
AI scoring no longer runs automatically when a workflow triggers a Pro AI translation task; it runs when the workflow includes it.
- Air — 5 new video features on Air
Caption, review, and ship your videos, all in one place.
- Stagehand — Cut inference costs 30–40% with Model Router
Set model: "auto", and Browserbase selects the model for each call server-side, balancing accuracy, speed, and cost automatically.
- Lumen5 — Transitions come alive in Lumen5
Lumen5 shipped a transitions library with options users can add between any two scenes.
- CapCut — CapCut AI Image Generator Preview: Why Seedream 5.0 Pro Matters for Design Studio Workflows
A preview explains Seedream 5.0 Pro in CapCut AI Image Generator for Design Studio workflows, campaign visuals, and creative iteration.
- CapCut — What to Expect from Seedream 5.0 Pro in CapCut's AI Image Generator
The article covers prompt control and campaign visual concepts for Seedream 5.0 Pro in CapCut's AI Image Generator.
- CapCut — Seedream 5.0 Pro and CapCut Design Studio: Faster Visual Creation for Creators
Seedream 5.0 Pro could support CapCut Design Studio workflows for posters, banners, social visuals, and campaign concepts.
- CapCut — How Seedream 5.0 Pro Could Upgrade AI Image Design in CapCut
Seedream 5.0 Pro could upgrade AI image design in CapCut with prompt control, references, and visual starts.
- CapCut — Seedream 5.0 Pro Is Coming to CapCut: A New Era for AI Image Generation
Seedream 5.0 Pro is coming to CapCut for AI image generation workflows for social visuals, campaigns, and design planning.
- Trae IDE — Create Tasks with Plugins on Mobile.
Added support for inserting plugins directly from the chat input when starting tasks from the App.
- Hyperscience Platform — 43.1.4 (27 Jul 2026)
Users can filter case documents by status; SAML-enabled upgrades use a valid default administrator-role setting.
- Trae IDE — Create Tasks with Plugins on Mobile
Added support for inserting plugins directly from the chat input when starting tasks from the App.
- Open WebUI — v0.11.0
Redesigned the interface and added sub-agents.
- Typefully — Publish and Schedule Substack Notes
Added the ability to create and schedule Substack Notes in Typefully.
- Clay — Product Roundup Week of Jul 20, 2026
Find enrichment providers with filterable badges; trace connection use and access; use higher field limits; track CPJ allowance; see near real-time credit spend.
- browser-use — 0.13.7
Shipped CLI updates with bug fixes and better browser management.
- Cerebras Inference — GLM 4.7 reasoning_logprobs default will not ship before deprecation
zai-glm-4.7 will keep its existing default logprobs response shape until deprecation. Requests that explicitly set the X-Cerebras-Version-Patch: 2 header still receive…
- ASI:One — Your AI Checks Its Own Work
After every answer, a second opinion looks for a tool that would have done better. When it finds one, you get a one-tap way to run it.
- Zencoder Coding Agent — July 2026
Added reasoning-effort selection and made Claude Fable 5, Sonnet 5, and GPT-5.6 available in Claude Code and OpenAI Codex.
- Zencoder Coding Agent — June 2026
Added provider API keys from .env files, global installation for Codex and Claude, updated Claude Opus, and credits usage visibility.
- Ludwig — v0.17.8: path traversal fix in dataset archive extraction
Fixed path traversal in dataset archive extraction through symlink members.
- Rossum — Document Comments
Add a free-text comment to any document, see the trail of notes, include comments when rejecting or postponing, and filter for documents with comments.
- Cerebras Inference — Image dimension limit for image inputs
Image inputs now enforce a maximum width and height of 15,000 pixels per side. Requests with larger images return 413 Content Too Large.
- Squibler — AI Generation on Large Documents
Fixed AI writing failures on longer documents when the context limit was reached; generation now handles large projects.
- Squibler — July 25th, 2026
Fixed AI generation on large documents, editor generation errors, real-time editing stability, dashboard reliability, project data accuracy, and user menu display.
- Squibler — Bug Fixes
Fixed AI writing failures on longer documents, generation errors, connection leaks, dashboard crashes, deleted projects reappearing, and user menu display.
- Rask AI — Cleaner Dubbed Audio: Background Separation and Even Volume
Background Separation and Even Volume
- Emergent — From Prompt to Production: Why We Rebuilt Our Deployment Engine that wasn’t on fire
We rebuilt our deployment engine.
- Culture Amp AI Coach — Expanded Language Support for the Performance Culture Diagnostic Template
PCD templates are translated across all Tier 2 languages; survey questions, communications, reporting, and action frameworks are localized for employees.
- Culture Amp AI Coach — AI Highlights and Opportunities in Probation Reviews
Managers can view up to 3 key highlights and 3 growth opportunities, with supporting feedback excerpts.
- Character.ai — Introducing Lorebook: Build the World Behind Your Characters
Character.ai launched Lorebook in beta, letting Characters draw on collections of world knowledge in chat; early access is for (c.ai)+ subscribers.
- Culture Amp Engage — AI Highlights and Opportunities in Probation Reviews
Managers can view up to 3 highlights and 3 growth opportunities with supporting feedback excerpts during manager reviews.
- Culture Amp Engage — Expanded Language Support for the Performance Culture Diagnostic Template
PCD templates are translated across 27 Tier 2 languages, including survey questions, communications, reporting, and action frameworks.
- Enterprise h2oGPTe — v1.7.3-post1
Adds Graph RAG, Guardrail Insights, an Agent Tool Server, and admin session management, along with reliability fixes.
- Cerebras Inference — Predicted Outputs limited to dedicated endpoints
Predicted Outputs is available by request on dedicated endpoints. You can also try Predicted Outputs with gpt-oss-120b in Public Preview.
- Claude AI — Claude Opus 5 launch
Launched Claude Opus 5, which comes close to the frontier intelligence of Claude Fable 5 at half the price.
- Midjourney — Version 8.2
The V8.2 image model launched with updates focused on aesthetics, image quality, and personalization.
- Altana Atlas — Week of July 23, 2026
Full-page entries keep header details; HS-code breadcrumbs show AI-generated descriptions; CSV uploads show filenames and use UTC timestamps.
- Flux — FLUX 3 - Real World Models: Towards Multimodal Flow Models as the Backbone of Visual Intelligence.
- Flux — FLUX 3 x mimic: The Next Generation of Video-Action Models