Superpower Daily
OpenAI Proposes November 12 End to Cursor Model Access After Sp...Anthropic Says Claude Found Fixes Across 10 Alignment Failures,...Lightwheel Releases 10,000 Hours of Human Video for Robots, Wit...Meta Tests Data-Center Repair Robots, but Cable Bots Still Need...SageMaker Feature Store Adds 25-Record Batch Writes and a Recor...JAMA Paper Says Autonomous AI Could Surpass Physicians and AI-A...OpenAI Proposes November 12 End to Cursor Model Access After Sp...Anthropic Says Claude Found Fixes Across 10 Alignment Failures,...Lightwheel Releases 10,000 Hours of Human Video for Robots, Wit...Meta Tests Data-Center Repair Robots, but Cable Bots Still Need...SageMaker Feature Store Adds 25-Record Batch Writes and a Recor...JAMA Paper Says Autonomous AI Could Surpass Physicians and AI-A...

Research report 09 / Developer cadence

Model Release Velocity

Release frequency, developer share, model availability, and cadence from verified model launches.

Archived snapshotv14Aug 29, 2026
View latest dataset
Verified releases
39

39 total model launches

Developers
27

Named release organizations

Most active
OpenAI

6 releases

Open share
33%

Disclosed open availability/license

Interactive figureRelease velocity
CSV JSON
Data status39 verified records across 7 periods

Each bar is one month. Colored segments show how verified observations are distributed across series.

Verified observationHover or focus any mark for exact valuesLast updated Aug 29, 2026

Version ledger

Frozen public editions

Each edition preserves the records, method, sources, and downloads available at publication time.

  1. v14 / latestAug 29, 202639 records / 40 sources

    +1 records; updated Verified releases, Open share. The frozen snapshot contains 39 records.

  2. v13Aug 28, 202638 records / 40 sources

    +1 records; updated Verified releases, Developers, Open share. The frozen snapshot contains 38 records.

  3. v12Aug 28, 202637 records / 40 sources

    +1 records; updated Verified releases, Open share. The frozen snapshot contains 37 records.

  4. v11Aug 28, 202636 records / 40 sources

    Source-backed fields changed; the frozen snapshot contains 36 records.

  5. v10Aug 28, 202636 records / 40 sources

    +1 records; updated Verified releases, Developers, Open share. The frozen snapshot contains 36 records.

  6. v9Aug 28, 202635 records / 40 sources

    +1 records; updated Verified releases, Developers, Open share. The frozen snapshot contains 35 records.

  7. v8Aug 27, 202634 records / 40 sources

    Source-backed fields changed; the frozen snapshot contains 34 records.

  8. v7Aug 27, 202634 records / 40 sources

    +1 records; updated Verified releases, Developers, Open share. The frozen snapshot contains 34 records.

  9. v6Aug 27, 202633 records / 40 sources

    Source-backed fields changed; the frozen snapshot contains 33 records.

  10. v5Aug 27, 202633 records / 40 sources

    Source-backed fields changed; the frozen snapshot contains 33 records.

  11. v4Aug 27, 202633 records / 40 sources

    Source-backed fields changed; the frozen snapshot contains 33 records.

  12. v3Aug 27, 202633 records / 40 sources

    Source-backed fields changed; the frozen snapshot contains 33 records.

  13. v2Aug 27, 202633 records / 40 sources

    Source-backed fields changed; the frozen snapshot contains 33 records.

  14. v1Aug 27, 202633 records / 40 sources

    Initial public snapshot with 33 records and 40 cited sources.

Coverage note

Release counts reflect the monitored source set and verified records, not every model published globally.

Dataset ID
spd:release-velocity
Stable URL
/research/release-velocity
Version
v14
Coverage
2024-10/2026-08-28
Records
39
Fields
6
Updated

Read the data

The records behind the figure

CSV JSON
Model Release Velocity data records
DateModelDeveloperRelease typeAvailabilityLicense
2024-10-07Inflection 3.0Inflection AIcommercial API and enterprise AI systemCommercial API available to developers; enterprise deployment on-premises, in the cloud, or hybrid—
2025-08-22GPT-4b microOpenAIexperimental researchdeveloped for research purposes; not broadly available—
2025-12-11GPT-5.2OpenAIAPI and ChatGPT (model variants: GPT-5.2 Instant, GPT-5.2 Thinking, GPT-5.2 Pro)In ChatGPT: GPT-5.2 Instant, Thinking, and Pro begin rolling out today to paid plans; In the API: available now to all developers.—
2025-12-16GPT Image 1.5OpenAIAPI and ChatGPT rolloutModel rolling out in ChatGPT for all users and available in the API as GPT Image 1.5; the new Images experience is rolling out to most ChatGPT users, with Business and Enterprise access coming later; global.—
2026-04-07Claude Mythos PreviewAnthropicPreviewrestricted access (restricted-access rollout)—
2026-04-28Laguna XS.2Poolsideopen weightsPreview; Hugging Face, Poolside API, and OpenRouterApache 2.0
2026-06-09Claude Fable 5AnthropicPublic release (Mythos-class); available via API and EnterprisePublicly accessible via API and Enterprise; included in subscription plans through 2026-06-22 after which access requires purchase of usage credits; some sensitive-topic access restricted and Mythos 5 limited to vetted Project Glasswing participants—
2026-06-30Claude Sonnet 5Anthropicgeneral availability and APIAll Claude plans and Claude API—
2026-07-02Laguna XS 2.1Poolsideopen weightsPoolside API, OpenRouter, Hugging Face weights, and local deploymentOpenMDW-1.1
2026-07-08GPT-LiveOpenAIvoice model (full-duplex)Rolling out to ChatGPT users globally (ChatGPT Voice); API availability planned/coming soon. (Update 2026-07-31: SynthID watermarking and API verification access added for supported audio.)—
2026-07-09GPT-5.6 (Sol, Terra, Luna)OpenAIgeneral availabilityGeneral availability following a limited preview—
2026-07-16Kimi K3Moonshotopen-weights release (weights published for public download)Weights published for anyone to download; model can be run locally on user hardwareLicense requires companies operating a model-as-a-service business with > $20M total annual revenue (over 12 months) to enter a separate agreement with Moonshot before commercial use.
2026-07-24Claude Opus 5Anthropicgeneral availabilityAvailable immediately; default on Claude Max and strongest model on Claude Pro—
2026-07-31MiniMax H3MiniMaxAPI and open weightsHosted API globally; downloadable weights restricted to the license's applicable territoriesMiniMax H3 Community License Agreement
2026-08-10Muse GlimmerMetaopen weightsConsumer hardware with 24 GB or 32 GB of memoryApache 2.0
2026-08-11Nemotron 3.5 LightningNvidiaopen weights (public release)Available through Nvidia and model repositories including Hugging Face and ModelScope; NeMo Switchyard available on GitHubOpenMDW-1.1
2026-08-12Qwen3.8-27B-CRACK-GGUFJinho Jang (dealign.ai)Open weights; GGUF quantizations packaged for local inference via llama.cppHugging Face repository; downloadable GGUF files and a 0.9 GB vision projector for local use—
2026-08-13GLM-5.3Z.aiAPITogether AI serverless and dedicated infrastructure—
2026-08-16GLM-5.3Zhipu (Z.ai)———
2026-08-18GPT-5.6 SolOpenAIAPILimited release to a small group of customers; wider access planned later—
2026-08-19Ornith-1.5Ornith AIopen weightsHugging Face collection; local deployment and quantized mobile version described—
2026-08-21V4-Flash-Vision-ExpDeepSeekexperimental API modelDeepSeek API—
2026-08-21deepseek-v4-flash-vision-expDeepSeekexperimental API modelAvailable through Chat Completions, Messages, and Responses requests—
2026-08-21Falcon 2Murf AIproprietary foundation voice model——
2026-08-24Thomson-1.0-LargeThomson Reutersproprietary model launch——
2026-08-24HiDream-O1-WorldHiDream.aimodel launch——
2026-08-24ThomsonThomson Reutersenterprise launch; open-weight version plannedInitially available in CoCounsel Legal’s Tabular Analysis feature; smaller version planned for Hugging FaceNon-commercial license for the planned smaller open-weight version
2026-08-24GEN-1.5Generalist AIresearch releaseAvailable through direct partnerships on Generalist AI's own robot fleet; no public weights, API, or self-serve product—
2026-08-26Qwen3.8-FlashAlibaba QwenAPIQwenCloud managed API—
2026-08-26Granite 4.2IBMopen weightsHugging Face, Ollama, GitHub, and other platformsApache 2.0
2026-08-26Qwen3.8-FlashAlibabaAPIAPI access—
2026-08-26Grok 4.6xAIpublic previewMicrosoft Foundry for Azure customers—
2026-08-26Qwen3.8-Flash-NextAlibaba Qwen teamarchitecture preview; open weights; production APIWeights on Hugging Face and ModelScope; production Qwen3.8-Flash through QwenCloud; API expected shortly—
2026-08-27GLM-5.3-FlashZ.aiAPI and open weightsZ.ai API; weights available on Hugging FaceMIT
2026-08-27GLM-5.3-FlashZhipu AIopen-sourceOfficially released; deployable on domestically produced AI chipsMIT
2026-08-27Gemini 3.5 TranscribeGoogleAPIPublic preview—
2026-08-27leiolai-1Leiolaiconsumer app and developer APIConsumer app and developer API—
2026-08-28ABot-ReconAMAP CV Labopen-source releaseInference code, model checkpoint, and evaluation code released; weights limited to noncommercial research and educationApache 2.0 for source code; CC BY-NC 4.0 for model weights
2026-08-28Sahara v2.5Intronmodel update——

Measurement technique

How to read this report

  1. 01Each verified model-launch record contributes one release observation.
  2. 02Developer names are normalized during extraction; series preserve currently disclosed attribution.
  3. 03Release volume measures activity, not capability or adoption.

Sources

Evidence

26 publishers supporting 40 records. Expand a publisher to inspect its cited pages.

the-decoder.comthe-decoder.com5 sources / 5 records
Thomson Reuters bets $40M on owning its AI instead of renting from OpenAI or Anthropic1 supporting recordIBM drops open-weight Granite 4.2 family with built-in agentic capabilities under Apache 2.01 supporting recordAlibaba releases Qwen3.8-Flash-Next, targeting "ultimate cost efficiency"1 supporting recordGoogle's Gemini 3.5 Transcribe turns speech to text in 85 languages while auto-correcting your verbal stumbles1 supporting recordGLM-5.3-Flash matches top models at a fraction of the cost, and runs without Nvidia1 supporting record
openai.comopenai.com4 sources / 4 records
GPT-5.6: Frontier intelligence that scales with your ambition1 supporting recordAccelerating life sciences research1 supporting recordIntroducing GPT-Live1 supporting recordThe new ChatGPT Images is here1 supporting record
Ars Technicaarstechnica.com3 sources / 3 records
IBM's new Granite 4.2 models ride the wave of interest in local LLMs1 supporting recordGoogle announces Gemini 3.5 Transcribe for AI-powered speech-to-text1 supporting recordAnthropic says these topics are too dangerous to let its Fable 5 model talk about1 supporting record
marktechpost.commarktechpost.com3 sources / 3 records
Generalist AI Releases GEN-1.5: A Robot Foundation Model That Learns New Tasks From One 3–12 Second Demo1 supporting recordAlibaba's Qwen Team Releases Qwen3.8-Flash-Next: A 125B Multimodal MoE With 6B Active Parameters Previewing the Qwen4 Architecture1 supporting recordZ.ai Releases GLM-5.3-Flash: A 320B-A18B Natively Multimodal MoE With a 1M-Token Context1 supporting record
campustechnology.comcampustechnology.com2 sources / 2 records
Meta Releases Muse Glimmer, an Open-Weight AI Model Designed to Run on Consumer Hardware -- Campus Technology1 supporting recordMeta Releases Muse Glimmer, an Open-Weight AI Model Designed to Run on Consumer Hardware -- Campus Technology1 supporting record
siliconangle.comsiliconangle.com2 sources / 2 records
Thomson Reuters launches proprietary AI model for legal work - SiliconANGLE1 supporting recordZ.ai open-sources ‘Ox Alpha’ model as GLM-5.3-Flash - SiliconANGLE1 supporting record
Show 20 more publishers
x.aix.ai2 sources / 2 records
Introducing Grok 4.61 supporting recordGrok 4.6 on Amazon Bedrock1 supporting record
anthropic.comanthropic.com1 source / 1 record
Project Glasswing: Securing critical software for the AI era1 supporting record
artificialintelligencemadesimple.comartificialintelligencemadesimple.com1 source / 1 record
Anthropic's Claude Mythos Launch Is Built on Misinformation1 supporting record
autoclaw.z.aiautoclaw.z.ai1 source / 1 record
AutoClaw Models | GLM Model Support1 supporting record
blog.googleblog.google1 source / 1 record
Intelligent transcription with Gemini 3.5 Transcribe1 supporting record
CNBCcnbc.com1 source / 1 record
Z.ai shares surge 8% after releasing new AI model running only on Chinese chips1 supporting record
cnet.comcnet.com1 source / 1 record
The Powerful Stealth AI Model ‘Ox Alpha’ Is Now GLM-5.3-Flash, and You Can Use It - CNET1 supporting record
cryptobriefing.comcryptobriefing.com1 source / 1 record
Zhipu launches GLM-5.3-Flash, its first natively multimodal model built for Chinese chips1 supporting record
deccanchronicle.comdeccanchronicle.com1 source / 1 record
Alibaba Releases Smaller, Cost-Effective Qwen AI Model1 supporting record
economictimes.comeconomictimes.com1 source / 1 record
Alibaba's Qwen launches Qwen3.8-Flash AI model with lower training costs - The Economic Times1 supporting record
globaltimes.cnglobaltimes.cn1 source / 1 record
Zhipu AI releases GLM-5.3-Flash model, powered by 100,000 domestically made chips - Global Times1 supporting record
invezz.cominvezz.com1 source / 1 record
Alibaba unveils cheaper Qwen AI model: is this what its stock needed?1 supporting record
itbrief.newsitbrief.news1 source / 1 record
Leiolai launches AI model that runs on user devices1 supporting record
readthepeak.comreadthepeak.com1 source / 1 record
Thomson Reuters Launches Its Own AI Model1 supporting record
stocktitan.netstocktitan.net1 source / 1 record
Thomson Reuters launches its own AI model for legal document reviews1 supporting record
TechCrunchtechcrunch.com1 source / 1 record
techcrunch.com1 supporting record
technologyreview.comtechnologyreview.com1 source / 1 record
Inside OpenAI’s big play for science1 supporting record
thelogic.cothelogic.co1 source / 1 record
Thomson Reuters launches its own AI model to reduce reliance on big tech - The Logic1 supporting record
vir.com.vnvir.com.vn1 source / 1 record
HiDream.ai releases HiDream-O1-World interactive artificial intelligence model1 supporting record
yahoo.comyahoo.com1 source / 1 record
Alibaba's Qwen launches Qwen3.8-Flash AI model with lower training costs1 supporting record
Next report / 10AI Attention vs. Substance Index All research reports