Gemini Omni 1.1 Flash went GA on August 27, 2026 — three months later it's superseded by 1.2: from Dec 3, 1.1 API requests auto-route to 1.2 at 1.2 prices. The migration needs no code change and throws no error — but output comes from a different model, and the extension cap shrinks. This article covers what changed, what actually happens on Dec 3, and what to do before migrating.
Duration: generation and editing expanded from 3–10s to 3–15s (whole-second steps, default 10s). Price: video output fell from $17.50 to $12.00/1M tokens — exactly 31.4%, Google says 30%; 720p per-second from $0.101 to $0.07; 10s 720p from ~$1.01 to ~$0.70. Input and text-output prices are the same ($1.50/$9.00). Rate limits: the changelog says 1.2's paid limits are 2x 1.1's. Token caps are the same (input 131,072, output 65,536). Quality claims: faster generation, stronger reference adherence, better consistency, more natural dialogue, steadier camera motion, fewer jump cuts — all vendor claims, unverified by third parties.
1.1's extension: +10s per turn, 40s cumulative cap. 1.2's extension: +3–10s per turn (default 10s), 30s total cap. This is the easy-to-miss downgrade at migration: flows that depend on 30–40s total length can't reach 40s after migrating. Note the sourcing difference: 1.1's 40s comes from multiple reports of Google's Aug 27 announcement (the original announcement wasn't read directly); 1.2's 30s comes from the developer guide itself.
1.1 is kept for a 4-week transition; from Dec 3, 2026, all requests to gemini-omni-1.1-flash auto-route to gemini-omni-1.2-flash at 1.2 prices — no code change needed, no error, but output comes from a different model. The pricing page, developer guide and changelog all agree. Three notes: first, routing applies only to the Gemini API ID; on Google Cloud, 1.1's ID is gemini-omni-1.1-flash-preview, marked deprecated with 1.2 recommended, but no date given — don't write it as a "Dec 3 switch across all platforms". Second, Google hasn't said whether routed responses report 1.1 or 1.2 in the model field. Third, after Dec 3 the docs offer no way to keep getting 1.1 output. Dec 5 is Flow's 1.1 retirement date — don't confuse the two.
On the AA video board: 1.1 text-to-video 1127 (#12, Nov 9 read), 1.2 at 1154 (#6, Nov 12 read) — up from #12 to #6; image-to-video 1.1 was #17, 1.2 at 1086 (#14); editing 1.2 at 1164 (#2), #1 in both complex-editing and VFX-editing sub-items. Note AA recalibrates Elo periodically, so scores from different dates aren't directly comparable.
First, check extension flows: if a flow depends on 30–40s total length, the post-migration cap is 30s — rework the flow or split into segments. Second, update the model ID: the old ID auto-routes after Dec 3, but explicitly switching to gemini-omni-1.2-flash is safer — Google hasn't said what routed responses report in the model field, and integrations that use that field for billing or log attribution could be affected. Third, re-measure cost: video output fell 31.4%, but input and text-output prices are unchanged; RanThis measured $1.05 including thinking tokens for one 15s 720p job — budget for thinking tokens. Fourth, re-test quality: 1.2's output comes from a different model, so prompt behavior may change; a 360p draft and the same-prompt 720p are completely different scenes, so don't use 360p as a preview in workflows. Fifth, confirm regional limits: EEA, Switzerland and UK users can't edit or extend uploaded video — a restriction to confirm at migration time.
Reference: beri.net — Gemini Omni 1.2 Flash GA: 720p Pricing, December 3 Model Routing (rewritten and organized by ModelBrief; facts per the source)