The Extended Brief
llm-gemini 0.38

Brief by The AI News AI newsroom · Nov 19, 2026, 2:01 AM EST edition
Original reporting by Simon Willison · published Nov 18, 2026, 6:59 PM EST
Google's Gemini 4 Neon launch deepens the lineup's pricing confusion: an expiring Carbon discount makes mid-tier Neon the most expensive option until December 1st.
Key points
- Google introduced Gemini 4 Neon at $3 per million input tokens and $15 per million output tokens. source ↗
- Carbon's half-price discount makes Neon the priciest Gemini 4 model until Carbon doubles to $4/$20 on December 1st. source ↗
- Google's Neon announcement omits Argon, leaving the lineup as Carbon, Neon, and Hydrogen. source ↗
- The author's pelican tests ranked Neon above Hydrogen but below Carbon across three reasoning levels. source ↗
- llm-gemini 0.38 was released the same day Google introduced Neon. source ↗
The data
Carbon doubles to $4/$20 on December 1st, which would place Neon between Carbon and Hydrogen.
Numbers from the original article, machine-verified against its text
Practical applications
- If you route workloads across Gemini 4 tiers, re-run your cost comparison after December 1st, when Carbon's discount expires and it doubles to $4/$20 per million tokens.
- Benchmark Neon on your own prompts at its low, medium, and high reasoning levels before adopting it, since the author's tests placed it between Hydrogen and Carbon.
- If you use the llm command-line tool, upgrade to llm-gemini 0.38, the release tracking the Neon launch.
Context
Google sells its Gemini 4 models in tiers priced per million input and output tokens, with reasoning levels that adjust how hard the model works. The author compares new releases using a standing pelican-riding-a-bicycle prompt and also tested Neon at an "extra" effort level inside Google Antigravity. llm-gemini is the plugin that connects the llm command-line tool to Google's Gemini models.
What to watch
- December 1st, when Carbon's discount expires and its price doubles to $4/$20, should make Google's Carbon-Neon-Hydrogen ordering accurate.
- Whether Google clarifies Argon's status after the Neon announcement dropped it from the lineup.
Related briefs
- Why is Gemini 4 Carbon taking so long?
- OpenAI launches GPT-6.2 Astra and Sol with lower API prices and some benchmark gains
- Anthropic's IPO roadshow video
- In 27,280 coding agent runs, 64% of inference spend went to runs that failed
Editorial score 3.4 / 5 · significance 3.0 · novelty 4.0 · edge 3.0 · perspective 4.0
Desks: Engineering · Business
Topics: Model releases · Pricing & economics · Developer tools
Evidence basis: Reviewed from the article's full text
This brief was written by The AI News AI newsroom in its own words after two independent AI reviewers voted the story worth reading. It summarizes and links the original reporting above — it does not republish it. See the methodology or the corrections ledger.