LeaksinclusionAI
Ling 3.2 Flash spotted on OpenRouter
Ling 3.2 Flash from inclusionAI was first recorded on 13 Nov 2026. Its release date is not established. 2 evidence links.
Watching 36 AI labs around the clockupdated
Know about new AI models before they’re announced.
The tracker watches model catalogues, API listings, cloud quotas, arenas and code. When a new name shows up, it lands here with its sources and Alex’s take on what it means.
Coming soon
Unreleased models we have found, how close each one is to launch, and where it was last seen.
Sighting
GPT-6.2 Astra and GPT-6.2 Sol popped up in Google Cloud’s quota settings, stayed for…
46 dayson the radar
Sighting
GPT-6.2 Astra and GPT-6.2 Sol popped up in Google Cloud’s quota settings, stayed for…
9 dayson the radar
Reported
Claude Mythos 5.5 has been spotted on Anthropic’s official website It’s mentioned four…
31 dayson the radar
Reported
Claude Mythos 5.5 has been spotted on Anthropic’s official website It’s mentioned four…
12 dayson the radar
Reported
🚨 Opus 5.2 appeared in the Microsoft Foundry catalog At first glance it looks like a…
60 dayson the radar
Reported
#15762 fix(core): size the qwen4 generation at 1M
79 dayson the radar
GPT Astra · Timeline
Grouped by product family. These names may refer to different models.
Claude Mythos · Timeline
Grouped by product family. These names may refer to different models.
Claude Opus · Timeline
Grouped by product family. These names may refer to different models.
Findings with unknown purpose · 2
These identifiers were observed; their purpose and any future release are unconfirmed.
Reported only in posts or pressSighting in a catalogue, quota or codeAccess announced, limited or callableReleased official
We saw them coming
12 daysMedian time from our first mention to release, across 11 launches. Includes reports and rumours, not verified predictions. Claude Fable 5.2 was first mentioned 61 days before release.
How we measure it →- Claude Fable 5.2 → , 61 days
- GPT-6 Astra → , 27 days
- GPT-6 Sol → , 15 days
- GPT-6 Luna → , 15 days
- Gemini 4 Carbon → , 12 days
- Gemini 4 Argon → , 12 days
- Grok 4.7 → , 3 days
- Claude Opus 5.5 → , 1 day
Follow your tools
Releases, limits, prices and leaks for the AI you use every day, in one timeline each.
Assistants and their models
ChatGPT
Events: 87· Tools & updatesGPT Dots is starting to roll out across the EU.
Claude
Events: 85· Prices & limitsClaude just gave everyone a banked reset!
Gemini
Events: 31· Tests & comparisonsGemini 4 Hydrogen beats Argon in the Artificial Analysis intelligence index at the standard High reasoning level (51…
Kimi
Events: 12· Tests & comparisonsOpen-weight model Kimi K3.1 Pro ranks second on the Artificial Analysis leaderboard
Coding agents
Codex
Events: 72· Prices & limitsA free Codex reset is banked for everyone
Claude Code
Events: 15· Tools & updatesClaude Code just added idle compaction, and this could save a LOT of usage.
Cursor
Events: 6· Prices & limitsElon Musk just combined Cursor, Grok, and Twitter subscriptions into one.
OpenCode
Events: 6· Prices & limitsAI subscription buying guide:
Prices today
What the flagship models cost per million tokens; each model page cites the source of every rate.
| Model | Input | Output | Context window |
|---|---|---|---|
Claude Haiku 5.5 Anthropic | $0.10–$0.50 | $0.50–$2.50 | 1M |
GPT-6.1 Terra OpenAI | $1.00 | $5.00 | 1M |
Gemini 4 Hydrogen Google | $1.50 | $7.50 | 1M |
| Gemini 4 CarbonGoogle | $2.00 | $10.00 | 1M |
| GPT-6.1 SolOpenAI | $2.00 | $10.00 | 1M |
| Claude Fable 5.2Anthropic | $8.00 | $40.00 | 1M |
USD per 1M tokens. A range means the rate depends on prompt length or mode; the model page lists the conditions.
Alex’s take
What changed, why it matters, and what’s still unconfirmed.
Notes
Alex
AlexOperation JensenBack has officially begun
Sign the petition and help bring Jensen home. Elon Musk and many of my fellow bloggers have already signed. https://standwithjensen.com/#sign
Tools & updatesOpenAIAlex
OpenAIAlexGPT Dots is starting to roll out across the EU
For those unfamiliar, Dots is OpenAI’s proactive assistant, similar to Grok’s bot or a locally running Hermes agent. It can work on its own cloud computer or control your machines via Codex…
Prices & limitsOpenAIAlex
OpenAIAlexA free Codex reset is banked for everyone
Tests & comparisons
GoogleAlex
GoogleAlexGemini 4 Hydrogen beats Argon in the Artificial Analysis intelligence index at the standard High reasoning level (51 vs 50), and at the max Xhigh level it…
Analysis10 min read
Codex, Claude Code or Cursor: choose the workflow
Compare repository tasks, editor and CLI workflows, subscription prices and the cost of accepted work.
Be first to know.
Alex posts every major find in his channels, and the signal feed updates as soon as the tracker records one.