Gemini in November 2026: Omni 1.2 Flash, Gemini 4 Carbon, Hydrogen, and Neon
October 9 through November 20, 2026: Omni 1.2 Flash GA, Gemini 4 Carbon public access, Hydrogen and Neon launches, Flow tools, Tasks in Gemini, and API migration dates.
Scope
This post covers October 9 through November 20, 2026. It lists changes we could verify against Google’s own announcements. It is not exhaustive. For September through early October context, see the October 2026 update.
Omni 1.2 Flash
On November 5, Google made gemini-omni-1.2-flash generally available with 15-second native clips, multi-turn editing on the Interactions API, and paid-tier video output near $0.07 per second at 720p. Full limits, pricing tables, and deadlines are in the dedicated Omni 1.2 Flash announcement.
Gemini 4 Carbon, Hydrogen, and Neon
Three Gemini 4 models shipped to developers and subscribers in November:
| Model ID | Release (public) | Price (per 1M tokens) | Thinking | Who can use it (high level) |
|---|---|---|---|---|
gemini-4-carbon | 2026-11-10 | Intro through 2026-11-30: $2 in / $10 out; from 2026-12-01: $4 in / $20 out | low, medium (default), high | Paid Gemini API, Android Studio, Gemini for business, Ultra in Gemini app / AI Mode / Antigravity; Fairwind cyber program |
gemini-4-hydrogen | 2026-11-12 | $1.50 in / $7.50 out (flat across 1M context) | low, medium, high (default), xhigh | Free and paid AI Studio, Gemini API, Gemini for business, Gemini app (auto for free users), Search, Antigravity Pro/Ultra, Managed Agents default |
gemini-4-neon | 2026-11-18 | $3 in / $15 out; cached input $0.15 | Four levels incl. xhigh | Paid API in AI Studio / Android Studio; AI Pro and Ultra in Studio, Antigravity, Gemini app; Gemini for business; Tasks can run on Neon |
Gemini 4 Carbon
Announced October 28 for Fairwind defenders; public availability landed November 10. 1M-token unified context with long decode up to 1M tokens. Cached input is 95% off ($0.10 per 1M through November 30, $0.20 from December 1). Not on the free API tier. See the Carbon introduction for positioning on cybersecurity and long-horizon agents.
Gemini 4 Hydrogen
Hydrogen targets fast coding and agentic workflows, adapted from Carbon. Google compares it with Gemini 3.8 Flash. Google’s figures: 2.5x faster at default high thinking; Terminal-Bench 4.1 57% vs 18.2% on 3.8 Flash; fabricated claims cut by more than 5x. Artificial Analysis Intelligence Index 51 at high and 54 at xhigh per Google’s blog. Default for Managed Agents in Antigravity.
Gemini 4 Neon
Neon sits between Carbon (hardest long-horizon work) and Hydrogen (everyday speed). Google’s figures: Artificial Analysis Intelligence Index 61 at xhigh (#3 overall); Vals Index #2 (70.9%); Terminal-Bench 4.1 65.2%; at default high, 32% lower cost per task than Carbon per Google. Not on the free tier.
Flow and image models
On November 12, Google announced five Flow tools: Character Poster, Character Arc, Pixel Machine, Lucid Glass, and Pocket Studio.
Flow image defaults (per Flow help): Nano Banana Pro for Ultra, Nano Banana 2 Lite at no charge, Nano Banana 2.1 as standard. November 16 Workspace Updates moved Google Pics to Nano Banana 2.1 with sharper edits and text rendering, including Pics inside Docs and Slides (consumer rollout 1 to 3 days for Google AI Plus, Pro, and Ultra).
Video in Flow is now Omni Flash 1.2 and Veo 3.1 Lite; retirements are listed in the Omni 1.2 deadline table.
Gemini app and Gemini API
| Topic | Summary |
|---|---|
| Tasks in Gemini (blog, 2026-11-17) | Calls, bill pay, and purchases; free rollout in U.S., Japan, Korea, Canada, Australia, India on gemini-4-hydrogen; advanced features start in the U.S. |
| Mac computer use (blog, 2026-11-19) | Isolated background window on Mac; English, U.S.; Apple Silicon, macOS Sequoia 15.0+; Google AI Plus, Pro, or Ultra |
| Lower-memory Android (blog, 2026-11-18) | Gemini Live in 45+ languages, camera help, Nano Banana images on Go edition devices |
| Tool Retrieval (blog, 2026-11-19) | Preview in Interactions API: tool_search plus defer_loading: true; works with Gemini 3.6+ and all Gemini 4 models; no separate charge |
Paid-tier pricing for these models is on the AI Studio pricing page.
Deadlines and price changes
Consolidated dates that affect API and Flow users:
| Date | Surface | Change |
|---|---|---|
| 2026-11-30 | Gemini API | Gemini 4 Carbon introductory price ends |
| 2026-12-01 | Gemini API | Carbon becomes $4 input and $20 output per 1M tokens; gemini-2.5-pro routes to gemini-3.8-flash at 3.8 Flash rates |
| 2026-12-03 | Gemini API | gemini-omni-1.1-flash routes to gemini-omni-1.2-flash |
| 2026-12-05 | Google Flow | Omni Flash 1.1, Veo 3.1 Fast, and Veo 3.1 Quality retired |
| 2026-12-13 | Gemini Enterprise Agent Platform | gemini-live-2.5-flash-native-audio retired (replacement gemini-3.8-live) |
Additional lines from Google’s deprecations tables. The first two dates have already passed:
| Date | Surface | Change |
|---|---|---|
| 2026-11-17 | Gemini API | gemini-3.1-flash-tts-preview shutdown; use gemini-3.8-flash-tts or gemini-3.8-flash-lite-tts |
| 2026-11-19 | Gemini Enterprise Agent Platform | gemini-3.6-flash retirement (replacement gemini-3.8-flash) |
| 2027-01-28 | Gemini Enterprise Agent Platform | gemini-3.7-flash retirement (replacement gemini-3.8-flash) |
The October update recorded Google’s September guidance that Gemini 2.5 stays available until further notice. The December 1 gemini-2.5-pro reroute is a newer dated change on the Gemini API surface only.
November action checklist
- Migrate Omni video workloads to
gemini-omni-1.2-flashand budget near $0.07 per 720p second. - Before November 30, finish Carbon experiments on intro pricing if you need the lower rate.
- Evaluate
gemini-4-hydrogenfor agents andgemini-4-neonfor heavier software tasks; confirm tier access in AI Studio. - Confirm nothing still calls
gemini-3.1-flash-tts-previewon the Gemini API. It was scheduled to shut down on November 17. - Plan the December 1
gemini-2.5-protogemini-3.8-flashmigration on the Gemini API (review breaking changes in Google’s migration guide). - In Flow, prepare for December 5 model retirements and verify extension flows use Omni 1.2.