Digest · since Jul 3, 2026

What moved in the last 30 days

Every delta the daily crawl found across the tracked catalog — grouped by type, with the before and after values and a source link on each row.

version

36

pricing

7

model

14

capability

26

New to the tracker

Abridge

Abridge

Ambient clinical documentation

Version
Contextual Reasoning Engine
Cost
Enterprise contract pricing
Model
In-house medical models
AdCreative.ai

AdCreative.ai

Generated ad creative at scale

Version
2026 platform
Cost
Starter from $39/mo
Model
Undisclosed image models

Commercially safe generative imaging inside Creative Cloud.

Version
Firefly 5
Cost
from $9.99/mo
Model
Firefly Image 5
Agentforce

Salesforce

Autonomous service and sales agents grounded in CRM data.

Version
Agentforce 3
Cost
from $2/mo
Model
xGen
AgentKit

OpenAI

Visual agent builder, connectors and evals from OpenAI.

Version
2025
Cost
API usage-based
Model
GPT-5
AgiBot

AgiBotChina

Chinese humanoid maker shipping units at volume.

Version
A2
Cost
Unit sales
Model
GO-1
AI/BI Genie

Databricks

Ask your lakehouse questions

Version
2026 release
Cost
Consumption-based DBU pricing
Model
Llama
Aider

Aider

Terminal-native pair programmer that edits your git repo.

Version
0.8x
Cost
Open source, model costs only
Model
Claude Sonnet
Aidoc

Aidoc

Radiology triage foundation models across modalities.

Version
CARS
Cost
Enterprise quoted
Model
CARS foundation model
ALIA CX300

BETA Technologies

Electric fixed-wing cargo and passenger aircraft

Version
ALIA CX300
Cost
Aircraft sales plus charging network access
Model
In-house flight control
Alice & Bob

Alice & BobEU

Cat-qubit approach to hardware-level error protection.

Version
Boson 4
Cost
Cloud + partnerships
Model
cat qubit
AlphaSense

AlphaSense

Market intelligence search

Version
Generative Search
Cost
Seat-based enterprise pricing
Model
Undisclosed frontier LLM

Managed multi-vendor quantum cloud

Version
GA
Cost
Free simulator tier, per-task hardware pricing
Model
IonQ / IQM / Rigetti / QuEra

Enterprise assistant grounded in a company's AWS-connected content.

Version
2025
Cost
from $3/mo
Model
Claude

Ambient documentation with coding accuracy focus.

Version
2025
Cost
Enterprise quoted
Model
OpenAI

High-memory GPU alternative

Version
MI355X
Cost
~$2-3 per GPU-hour on clouds
Model
CDNA 4
Amp

Sourcegraph

Agentic coding on top of large-codebase context.

Version
Amp
Cost
Usage-based credits
Model
Claude Sonnet 4.5
Apollo

Apptronik

General-purpose humanoid for logistics

Version
Apollo
Cost
Pilot deployments; RaaS pricing not public
Model
Gemini Robotics

On-device generative features across iPhone, iPad and Mac.

Version
iOS 26 era
Cost
Free with supported Apple devices
Model
Apple On-Device (~3B)

On-device inference across Mac, iPhone and iPad.

Version
M5
Cost
Bundled with hardware
Model
Apple Foundation Models
AssemblyAI

AssemblyAI

Speech intelligence API

Version
Universal-2
Cost
Pay as you go from $0.12/hr
Model
Universal-2
Atlas

Boston Dynamics

Electric humanoid for industrial work

Version
Electric Atlas
Cost
Pilot deployments
Model
Large Behavior Models

Search, chat and agents across Jira, Confluence and connected SaaS.

Version
2025
Cost
Included with paid Jira/Confluence plans
Model
Claude
Atom Computing

Atom Computing

Neutral-atom systems with 1,000+ qubit arrays.

Version
2025
Cost
Partnership-based
Model
neutral atom

83 changes

Full feed
  • model

    LovableLovable added Gemini 3.6 Flash to its model stack

    Claude Opus 4.5, GPT-5.2, Gemini 3 ProClaude Opus 4.5, GPT-5.2, Gemini 3 Pro, Gemini 3.6 Flash

    source
  • capability

    LovableNew capabilities: AI software engineer, Full-stack web apps, GitHub sync

    58

    source
  • model

    CodexCodex added GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna to its model stack

    GPT-5.2, GPT-5.2-CodexGPT-5.2, GPT-5.2-Codex, GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna

    source
  • capability

    CodexNew capabilities: Codex CLI, Codex IDE extension, Codex cloud

    57

    source
  • version

    CursorCursor 2.4 ships with a faster Composer model and improved multi-root workspaces

    2.22.4

    source
  • model

    KimiModel lineup now K3, K2.7 Code (and high-speed variant) and K2.6; K2.5 and moonshot-v1 sunset for new users on Aug 31, 2026.

    Kimi K2.5, Kimi K2 ThinkingKimi K3, Kimi K2.7 Code, Kimi K2.6

    source
  • capability

    TraeSOLO autonomous agent mode opened to free users

    Builder modeBuilder mode, SOLO agent

    source
  • capability

    Qwen ChatDeep research mode opened to all users

    Web search, Image generationWeb search, Image generation, Deep research

    source
  • version

    Claude CodeClaude Code 2.1 introduces plan mode improvements and faster subagents

    2.02.1

    source
  • version

    KlingKling 2.6 improves motion coherence and native audio

    2.52.6

    source
  • version

    n8nn8n 1.130 adds native evaluations for AI workflows

    1.1201.130

    source
  • version

    DeepSeekDeepSeek V3.2 released with sparse attention and a cheaper long-context path

    V3.1V3.2

    source
  • pricing

    DeepSeekAPI input price cut by roughly half after the V3.2 efficiency work

    0.560.28

    source
  • model

    CursorAdded Gemini 3 Pro to the model picker

    Claude Opus 4.5, GPT-5.2, ComposerClaude Opus 4.5, GPT-5.2, Composer, Gemini 3 Pro

    source
  • version

    Z.ai GLMGLM-4.7 released with open weights under MIT

    GLM-4.6GLM-4.7

    source
  • capability

    PerplexityComet browser opened to all Pro subscribers

    Cited answers, Deep research, SpacesCited answers, Deep research, Comet browser, Spaces, Shopping

    source
  • version

    HailuoHailuo 02.5 raises default output to 1080p

    0202.5

    source
  • model

    GitHub CopilotDefault chat model moved to GPT-5.2

    GPT-5GPT-5.2

    source
  • model

    Gemini Code AssistCode Assist upgraded to Gemini 3 Pro with agent mode

    Gemini 2.5 ProGemini 3 Pro

    source
  • version

    KimiK2.5 ships with stronger agentic tool use and vision

    K2K2.5

    source
  • version

    WindsurfWindsurf 1.14 released with a rebuilt Cascade planner

    1.101.14

    source
  • capability

    Skydio X10Remote Ops dock updates enable fully unattended patrol missions

    Unattended dock patrols

  • version

    DoubaoDoubao 1.8 Pro rolled out across the consumer app

    1.61.8

    source
  • version

    Nano BananaNano Banana Pro released with 4K output and better text rendering

    Nano BananaNano Banana Pro

    source
  • model

    Qwen ChatQwen3-Max became the default model in chat

    Qwen3-235BQwen3-Max

    source
  • capability

    CursorBugbot automated code review moved out of preview

    Background agentsBackground agents, Bugbot review

    source
  • capability

    Microsoft 365 CopilotAnthropic models added alongside OpenAI in Copilot and Copilot Studio

    Multi-provider routing

    source
  • version

    Google TPU IronwoodIronwood (v7) TPUs reach general availability

    v5pv7 Ironwood

    source
  • version

    MiniMaxM2.1 released with interleaved thinking traces

    M2M2.1

    source
  • capability

    CodexCodex gained parallel cloud sandboxes and pull request drafting

    Cloud tasksCloud sandboxes, Parallel tasks, Pull request drafting

    source
  • model

    PerplexityClaude Opus 4.5 and Gemini 3 Pro added to the model picker

    Sonar, GPT-5Sonar, GPT-5.2, Claude Opus 4.5, Gemini 3 Pro

    source
  • version

    KimiMoonshot AI released Kimi K3, a 2.8T-parameter open-weight native multimodal agentic model.

    K2.5K3

    source
  • capability

    KimiContext window expanded from 256K to 1M tokens with K3.

    256K tokens1M tokens

    source
  • capability

    KlingElements multi-reference conditioning added

    Motion brush, Lip syncMotion brush, Lip sync, Elements references

    source
  • model

    Claude CodeOpus 4.5 became the default model for Max subscribers

    Claude Sonnet 4.5Claude Opus 4.5

    source
  • capability

    Waymo DriverFreeway and airport driverless service opened in additional metros

    Freeway + airport rides

  • version

    SeedreamSeedream 4.5 adds 4K output and better text rendering

    4.04.5

    source
  • version

    GPT ImageGPT Image 1.5 released with better text rendering

    GPT Image 1GPT Image 1.5

    source
  • version

    Isaac GR00TGR00T N1.6 released with Cosmos Reason integration

    N1.5N1.6

    source
  • capability

    KimiKimi Researcher moved out of invite-only preview

    Agentic tool useAgentic tool use, Kimi Researcher

    source
  • model

    WindsurfIn-house SWE-1.5 replaces SWE-1 as the default

    SWE-1SWE-1.5

    source
  • version

    ManusManus 1.5 doubles task length limits and adds wide research

    1.01.5

    source
  • capability

    MidnightPiloted transition flights completed as FAA type certification testing advances

    Full wing-borne transition

  • capability

    Figma AIFigma Make gains code layers and design-system awareness

    Make code layers

    source
  • version

    ERNIE BotERNIE 5.0 launched with native multimodal input

    4.55.0

    source
  • version

    ClaudeClaude Opus 4.5 released as the new flagship

    Opus 4.1Opus 4.5

    source
  • model

    Julius AIAdded Claude alongside GPT for analysis tasks

    GPT-5GPT-5.2, Claude

    source
  • model

    HunyuanHunyuan3D added to the assistant's generation menu

    Hunyuan-T2, HunyuanVideoHunyuan-T2, HunyuanVideo, Hunyuan3D

    source
  • version

    FLUXFLUX.2 released with open dev weights alongside the hosted pro model

    FLUX.1FLUX.2

    source
  • version

    IBM Quantum PlatformNighthawk processor announced alongside Heron r2 fleet expansion

    Heron r2Heron r2 / Nighthawk

    source
  • pricing

    TraePro tier repriced to $3/mo in most markets

    103

    source
  • version

    LovableLovable 2.0 adds agent mode and a rebuilt cloud backend

    1.92.0

    source
  • model

    Snowflake CortexClaude models made available natively inside Cortex

    Llama, Mistral, ArcticClaude, Llama, Mistral, Arctic

    source
  • capability

    Zipline Platform 2Platform 2 delivery radius expanded to new US metro partners

    New metro launches

  • version

    ServiceNow AI AgentsZurich release ships AI Agent Orchestrator and AI Control Tower

    YokohamaZurich

    source
  • pricing

    Groq LPUToken pricing cut across the open-model catalogue

    0.080.05

    source
  • version

    ViduVidu Q2 improves multi-reference consistency

    Q1Q2

    source
  • capability

    GitHub CopilotModel Context Protocol support became generally available

    Agent modeAgent mode, MCP support

    source
  • version

    ExaExa 2.0 released with a rebuilt neural index

    1.52.0

    source
  • pricing

    Z.ai GLMCoding Lite plan introduced at $3/mo

    153

    source
  • version

    MidjourneyVersion 7.2 improves prompt adherence and adds the new editor

    v7v7.2

    source
  • pricing

    ClayStarter plan raised from $134 to $149 per month

    134.00149.00

    source
  • pricing

    Unitree G1G1 base configuration price cut

    1690016000

    source
  • version

    AutoGLMAutoGLM 2.0 adds scheduled recurring tasks

    1.52.0

    source
  • pricing

    ChatGPTNew ChatGPT Go tier introduced at $8 per month

    Free, Plus $20, Pro $200Free, Go $8, Plus $20, Pro $200

    source
  • capability

    dbt CopilotFusion engine integration added to dbt Copilot

    Model generation, Test generationModel generation, Test generation, Fusion engine

    source
  • capability

    Walker S2Autonomous battery swap enables continuous multi-shift operation

    Self battery swap

  • version

    Adobe FireflyFirefly Image 5 raises native resolution and adds layered editing

    Firefly Image 4Firefly Image 5

    source
  • version

    Qwen-ImageQwen-Image-2 released under Apache 2.0

    Qwen-ImageQwen-Image-2

    source
  • version

    VeoVeo 3.1 adds scene extension and improved audio

    Veo 3Veo 3.1

    source
  • capability

    HexMagic became aware of dbt semantic models

    Magic AIMagic AI, Semantic model awareness

    source
  • capability

    ClaudeSkills and MCP connectors added to the consumer app

    Artifacts, ProjectsArtifacts, Projects, Skills, MCP connectors

    source
  • model

    HarveyAdded Gemini to the model panel alongside OpenAI and Anthropic

    GPT-5.2, ClaudeGPT-5.2, Claude, Gemini

    source
  • capability

    Joby S4Passenger-carrying demonstration flights begin in Dubai

    Piloted passenger demos

  • capability

    Le ChatLe Chat adds enterprise connectors with EU data residency

    Connectors GA

    source
  • version

    Figure 03Figure 03 unveiled with redesigned hands and wireless charging

    Figure 02Figure 03

    source
  • capability

    MiniMax SpeechEmotion control exposed in the speech API

    Voice cloning, 30+ languagesVoice cloning, 30+ languages, Emotion control

    source
  • version

    CursorCursor 2.2 adds background agents and extends context to 1M tokens

    2.02.2

    source
  • version

    GeminiGemini 3 Pro rolled out across the consumer app

    Gemini 2.5 ProGemini 3 Pro

    source
  • capability

    Cerebras WSE-3New throughput record published on a large open model

    record token/s

    source
  • version

    RunwayGen-4.5 released with stronger physics and motion consistency

    Gen-4Gen-4.5

    source
  • capability

    Palantir AIPAIP agents extended to edge and disconnected environments

    Edge agent runtime

    source
  • version

    Quantinuum HeliosHelios goes live, succeeding H2

    H2Helios

    source