On Nov 10 (UTC 16:00, 00:00 Beijing time Nov 11), Google's blog posted a short announcement: Gemini 4 Carbon is open to all developers. Two posts - one on the developer side (blog.google developer announcement), one on the model side (blog.google model announcement), both dated 2026-11-10.
The key word is "open." At its Oct 28 debut it was only released to Fairwind Program cyber defenders, and ordinary developers couldn't get it; this time the original reads "Today, we're making Gemini 4 Carbon available to developers, enterprises, scientists, and consumers." - all four gates opened at once. This answers the question left by our Oct 29 article: then it was defenders only, now developers are in.
Capability claims recorded as the official side states them: CursorBench 4.0, Perplexity WANDR, and Zapier AutomationBench all rank first per Google; third-party Artificial Analysis gives it 60, below Claude Fable 5.2’s 65 on Nov 9. All vendor-reported or third-party rankings; we don't endorse them.
The announcement gives two prices - the part to remember. Verbatim: now billed at "$2 per million input tokens and $10 per million output tokens," same as last-gen Argon; the footnote reads "After the introductory period expires on November 30, 2026, the price of $4 per 1M input tokens and $20 per 1M output tokens will apply." - after Nov 30, input and output both double.
| Tier | Input (per M tokens) | Output (per M tokens) |
|---|---|---|
| Carbon entry (through 11-30) | $2.00 ≈ ¥13.39 | $10.00 ≈ ¥66.95 |
| Carbon regular (from 12-01) | $4.00 ≈ ¥26.78 | $20.00 ≈ ¥133.90 |
| Our plaza gemini-3.1-pro (buyer 3.3 off) | $0.66 ≈ ¥4.42 | $3.96 ≈ ¥26.51 |
Conversion: official USD × 6.695 to CNY. At entry price, Carbon's input is ~3.0x our plaza's gemini-3.1-pro and output ~2.5x; post-hike ~6.1x and ~5.1x. It's flagship pricing, not a cheap model - not worth it for everyday high-volume calls.

At 09:40 on 2026-11-11, we pulled /api/v1/model-plaza and cross-checked the page: the plaza showed 111 in-stock models, 158 listings; searching carbon was empty with "no matching models"; argon was empty too. Calling directly with a clean-quota key, gemini-4-carbon and gemini-4-argon both returned 404 model_not_found.

To be precise: gemini-4-argon was on our plaza on Oct 29 at ~¥1.33/¥6.63, but today's search doesn't find it. That's a change on our catalog side, not a Google shutdown - both generations are on sale on Google's site. The only thing we can confirm is "you can't buy Carbon here right now."
gemini-4-carbon, same structure as Argon, switching is one string.gemini-3.1-pro has 1 seller at 3.3 off for buyers, ¥4.42 in / ¥26.51 out per M tokens - covers the vast majority of text and coding.Official info from the two blog.google announcements (2026-11-10), prices and dates verbatim, capability marked as vendor-reported or third-party; RMB conversion at 1 USD = 6.695, same as our token billing article. Our availability from the /api/v1/model-plaza API and page test (2026-11-11 09:40), verified 404 with a real call. This article did not test Carbon's quality or speed - no supply means no test, and no "who's stronger" comparison. Images are our screenshots and self-made.
GPT-6.2 cuts list prices 20%, yet platform discounts run deeper: gpt-6-astra trades at under 1% of list. What this week's AI price war means for buyer
Prompt caching stores a repeated prompt prefix and bills cache reads at 5-12% of normal input. Real prices, hit conditions, and how to cut your API bi
Twelve Claude versions are on sale. The newest opus-5-5 costs $8.91 a month; the oldest opus-4-6 costs $49.50. Full table and a pick for each workload
gpt-5.6-terra and deepseek-v4.1-flash win on price and reliability; gemini-3.8-flash is fastest. Effective prices, measured speed, and a three-tier pi
Point your base URL at the platform and one key covers all 138 on-sale models across 51 vendors, in both OpenAI and Anthropic request formats. Copy-pa