Mission Squad
All pages
Workflow On demand

Model Release Radar

A live radar sweep of newly released and updated AI models across the industry — OpenAI, Anthropic, Google, Meta, xAI, DeepSeek, Qwen, Kimi, Mistral, and dozens of other labs. Run it for the whole landscape or focus on a single lab.

by demo Powered by Model Release Radar workflow Last produced August 5, 2026 · 6:34 AM

Run this page

JavaScript is required to run this page

Every run is saved to history and searchable

The global news feed (GDELT) failed due to a 429 rate-limit error, but web search results reveal an exceptionally

all major AI labs

Produced August 5, 2026 · 6:34 AM

The global news feed (GDELT) failed due to a 429 rate-limit error, but web search results reveal an exceptionally active AI model release cycle. The landscape is marked by massive scale increases in both open-weights and proprietary models, led by Moonshot AI's release of the 2.8-trillion-parameter Kimi K3 open-weights model and Thinking Machines Lab's 975-billion-parameter Inkling. The balance of releases is highly competitive, spanning major proprietary API updates and massive open-weights weights from both US and Chinese labs.

up
Releases tracked 7 across 5 distinct labs
Open-weights releases 2 including 2.8T Kimi K3
Proprietary releases 3 including GPT-5.6 Sol
Labs shipping 5 US and Chinese labs active
Largest open model 2.8T Kimi K3 parameters

Moonshot AI — Kimi K3

Moonshot AI released the open weights of Kimi K3, a 2.8-trillion-parameter Mixture-of-Experts model, on July 27, 2026. It features a 1M-token context window and is released under a Modified MIT license, making it the world's largest open-weight model. It shows particular strength in coding benchmarks like SWE Marathon.

Thinking Machines Lab — Inkling

Founded by former OpenAI CTO Mira Murati, Thinking Machines Lab released 'Inkling' on July 16, 2026. Inkling is a 975-billion-parameter open-weights frontier-class model designed to offer an open alternative to proprietary frontier systems.

Alibaba (Qwen) — Qwen 3.8 Max

Alibaba previewed and launched Qwen 3.8 Max, a 2.4-trillion-parameter flagship model (with ~95B active parameters) and a 1M context window. Alibaba claims the model's performance is second only to Anthropic's Claude Fable 5.

OpenAI — GPT-5.6 (Sol)

OpenAI released GPT-5.6 Sol, a proprietary frontier model designed to scale intelligence and efficiency. The model achieves state-of-the-art results across coding, cybersecurity, and scientific tasks.

DeepSeek — DeepSeek V4

DeepSeek transitioned DeepSeek V4 to general availability on July 24, 2026. The proprietary model features a 1M-token context window and introduces a peak/off-peak pricing structure.

ModelLabAccessDateNote
Qwen 3.8 Max Alibaba (Qwen) proprietary 2026-08-03 2.4T parameter flagship with 1M context, claimed second only to Claude Fable 5.
Qwen-Image-2.0 Alibaba (Qwen) unclear 2026-08-01 Next-gen 7B image model with native 2K resolution and image editing capabilities.
DeepSeek-V4-Flash-0731 DeepSeek unclear 2026-07-31 Flash variant of DeepSeek V4 released at the end of July.
Kimi K3 Moonshot AI open weights 2026-07-27 Massive 2.8T parameter MoE model with 1M context and Modified MIT license.
DeepSeek V4 DeepSeek proprietary 2026-07-24 Reached GA with 1M context and peak/off-peak pricing.
Inkling Thinking Machines Lab open weights 2026-07-16 975B parameter open-weights model from former OpenAI CTO Mira Murati.
GPT-5.6 (Sol) OpenAI proprietary 2026-07-16 Frontier-class model scaling intelligence and efficiency.

Alibaba's Qwen team has promised to release the open weights for Qwen 3.8 Max within days of its August 3 launch. Additionally, the rapid succession of releases from OpenAI, Anthropic, Google, and Meta in July suggests a highly competitive summer release cycle with more frontier updates expected.

Source openai.com ↗

Run history

1 run
The global news feed (GDELT) failed due to a 429 rate-limit error, but web search results reveal an exceptionally all major AI labs up August 5, 2026