# Radar, November 19, 2026: ten signals from X on Gemini 4 Neon, Clairic and Grok Bot

_Thursday November 19, 2026 · AI provenance: This page was compiled by Claude (AI, Anthropic) from X search briefings by Grok Bot (AI, xAI) and the llmnet team's X trend monitor. AI wrote the text; every signal has a source link to X and a confidence level. Responsible for llmnet.nl: Ivo Donker._

**In short:** The key signal of November 19: Google released Gemini 4 Neon, the middle Gemini 4 model at 3 and 15 dollars per million tokens. Alongside it, ten short signals on Cloudflare's open extraction models, Grok Bot, Anthropic's IPO, Cloudflare's cf CLI and new agent APIs, each with its source on X.

## 1. Google releases Gemini 4 Neon: 3 and 15 dollars, 1 million context

Neon is the middle Gemini 4 model, between Carbon and Hydrogen, aimed at coding and business workflows. Google's blog lists 3 dollars per million input tokens, 0.15 dollars cached and 15 dollars output, and a 32 percent lower cost per task than Carbon. It is in the Gemini API, AI Studio, Antigravity and the Gemini app for Pro and Ultra. The benchmark figures come from Google.

Source on X: [@OfficialLoganK](https://x.com/OfficialLoganK/status/2123167791922741362) · [Google, Introducing Gemini 4 Neon](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-neon/)
Context in the briefing: [Gemini 4 Neon in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#gemini-4-neon)
Read also: [Gemini 4 Neon](https://nieuws.llmnet.nl/en/gemini-4-neon) · [Gemini 4 Carbon](https://nieuws.llmnet.nl/en/gemini-4-carbon)

## 2. Two CursorBench figures for Neon: 63.7 and 65.4 percent

Cursor reports 63.7 percent on CursorBench 4.1 for Neon at the default setting. The 09:00 radar briefing reads 65.4 percent in Google's model card, with no setting stated. Two measurement settings, two numbers; we did not read the model card ourselves. Do not treat one figure against the other as evidence of progress.

Source on X: [@cursor_ai](https://x.com/cursor_ai/status/2123198346802450926)
Context in the briefing: [Gemini 4 Neon in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#gemini-4-neon)
Read also: [Gemini 4 Neon](https://nieuws.llmnet.nl/en/gemini-4-neon)

## 3. Cloudflare makes Clairic (27B) and Clairic-flash (9B) open-weight

The two extraction models pull fields from text, images and now PDFs, up to four pages per request. On Hugging Face they are under Apache 2.0, based on Qwen3.8-27B and Qwen3.5-9B. On Workers AI they cost 0.10 dollars per million input tokens until December 1; after that Clairic costs 0.24 dollars. Anyone on the preview endpoint gets the larger model automatically.

Source on X: [@corywilkerson](https://x.com/corywilkerson/status/2123182780070949275) · [Cloudflare, open Clairic](https://blog.cloudflare.com/open-clairic)
Context in the briefing: [Cloudflare in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#cloudflare-clairic)
Read also: [Clairic open-weight](https://nieuws.llmnet.nl/en/cloudflare-clairic-open-weights)

## 4. Anthropic prices its IPO at 83 dollars per share

Anthropic's account announces the IPO pricing; press sources put it at 83 dollars per share, above the earlier 70 to 78 dollar range, under the ticker ANT. Proceeds and valuation differ by source and are not in the announcement itself. Trading starts on November 19, according to the reports.

Source on X: [@AnthropicAI](https://x.com/AnthropicAI/status/2123196808675405860) · [StockAnalysis](https://stockanalysis.com/news/anthropic-prices-ipo-83-per-share/)
Context in the briefing: [The IPO in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#anthropic-ipo)
Read also: [Anthropic's S-1/A](https://nieuws.llmnet.nl/en/anthropic-s1a-q3-2026)

## 5. Grok Bot books flights, hotels and rental cars, according to @bot

The Grok Bot account says Bots can now book travel. The November 11 Grok Bot 1.0 announcement, which we read, covers Team Bots for Team and Enterprise and meetings for Team, Enterprise and Ultra, but no bookings yet. How payment and confirmation work is not in anything we read. Set a spending limit before connecting it.

Source on X: [@bot](https://x.com/bot/status/2123121401762918718) · [xAI, Introducing Grok Bot 1.0](https://x.ai/news/grok-bot-1-0)
Context in the briefing: [Grok Bot in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#grok-bot-reizen)
Read also: [Grok Bot 1.0](https://nieuws.llmnet.nl/en/grok-bot-1-0-team-bots)

## 6. Cloudflare: a cf CLI for the whole API, and WAF rules against Java4Shell

Cloudflare's developer account announces a release candidate of cf, one command-line tool for the whole public Cloudflare API, with cf migrate for Wrangler projects. Cloudflare also reports Java4Shell, a critical memory flaw in OpenJDK's Base64 decoder, with six WAF rules for paid plans. We did not read the OpenJDK advisory; check your Java versions with your vendor.

Source on X: [@CloudflareDev](https://x.com/CloudflareDev/status/2123094830045921764) · [@Cloudflare](https://x.com/Cloudflare/status/2123067084976308267)
Context in the briefing: [Cloudflare in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#cloudflare-clairic)
Read also: [Clairic open-weight](https://nieuws.llmnet.nl/en/cloudflare-clairic-open-weights)

## 7. Report of prompt leakage in Claude Code 2.1.327 not yet confirmed

A GitHub issue reports a user seeing other users' requests in their VS Code chat window. A post on X says it was forwarded to HackerOne and has not been reproduced. We did not verify the issue ourselves. Until Anthropic responds, this is a single report.

Source on X: [@bebuu_ai](https://x.com/bebuu_ai/status/2123288867042476041) · [GitHub-issue 111061](https://github.com/anthropics/claude-code/issues/111061)
Context in the briefing: [Unconfirmed in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#onbevestigd)
Read also: [Claude Code 2.1.326](https://nieuws.llmnet.nl/en/claude-code-2-1-326-remote-control)

## 8. OpenAI background mode: 30 or 40 percent off for patient work

OpenAI's developer account highlights background mode in the Responses API: if you can wait two or eight hours for an answer, you pay 30 or 40 percent less. This was already in the November 17 changelog, which we read for our GPT-6.2 article. Useful for overnight evaluations and batches.

Source on X: [@OpenAIDevs](https://x.com/OpenAIDevs/status/2123188195202617408) · [OpenAI, API-changelog](https://developers.openai.com/api/docs/changelog)
Context in the briefing: [Agent APIs in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#agent-apis)
Read also: [GPT-6.2 Astra and Sol](https://nieuws.llmnet.nl/en/gpt-6-2-astra-sol)

## 9. Perplexity adds a Context API to Monitor and Orchestrator

Aravind Srinivas announces a Context API alongside the earlier Monitor and Orchestrator APIs; Perplexity brings Teammates and Multiagent to its mobile app. We did not read the Context API's documentation, prices or terms.

Source on X: [@AravSrinivas](https://x.com/AravSrinivas/status/2123085971625476277) · [@perplexity_ai](https://x.com/perplexity_ai/status/2123126865829515363)
Context in the briefing: [Agent APIs in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#agent-apis)
Read also: [Perplexity Orchestrator API](https://nieuws.llmnet.nl/en/perplexity-orchestrator-api)

## 10. Qwen4-Max free for five days in Nous Portal; Mistral Vibe on desktop

Nous Research makes Qwen4-Max free for five days in Nous Portal, with image input and a 1 million token context; the model stays API-only. Mistral announces Vibe for desktop and says Vibe with Le Chonk can be tried for free in Europe. We did not read terms or durations outside X.

Source on X: [@NousResearch](https://x.com/NousResearch/status/2123135537087181191) · [@MistralAI](https://x.com/MistralAI/status/2123073168659141119)
Context in the briefing: [Hermes, Qwen and Vibe in the November 19 briefing](https://trending.llmnet.nl/en/briefing/2026-11-19#hermes-qwen-vibe)
Read also: [Qwen4-Max](https://nieuws.llmnet.nl/en/qwen4-max-preview) · [Mistral Large 4 Le Chonk](https://nieuws.llmnet.nl/en/mistral-large-4-le-chonk) · [Hermes Agent 0.23.2](https://nieuws.llmnet.nl/en/hermes-agent-0-23-2-stable)

[← Previous radar](https://radar.llmnet.nl/en/dag/2026-11-18)
[Next radar →](https://radar.llmnet.nl/en/dag/2026-11-20)
