Operation Vivaldi began before the advance. It moved a sector, not the war

The finding

On 14 October 2026 Ukraine’s 3rd Army Corps announced that the fourth and last phase of an operation it called Vivaldi was over. Three more villages north of Lyman, in the northern part of Donetsk oblast, were back under Ukrainian control: Zelena Dolyna, Lypove and Kolodiazi. Across the four phases, the corps said, it had retaken 277 square kilometres and sixteen settlements. Five days later its commander, Andriy Biletskyi, declared the operation complete, with the line on a small river called the Chornyi Zherebets.

That is the visible result, and it is the part most reporting stopped at. It is also the last part of the operation. The corps had taken over this sector more than a year earlier, the operation itself had been running in secret since late May, and the ground began to move months before anyone outside the corps called it an operation. The advance was downstream of a longer process, and the process is what this Note explains.

It then asks a second, separate question. On a front where both armies have measured progress in square kilometres a month for two years, did Vivaldi’s movement become operational — did it change what the Russian force in the sector could do, beyond the ground it lost? And if it did, does that change the character of the eastern campaign? The answers are yes, locally; and no.

Three terms carry the argument. Attrition is a way of fighting in which each side tries to wear the other down faster than it is worn down itself; ground moves slowly and the measure of progress is cost. Operational movement is an advance whose effects reach past the ground taken — an objective the other side must abandon, reserves it must move, a line it can no longer hold. Manoeuvre at campaign scale is movement of that kind sustained across a front, so that the war itself stops being decided by attrition. Vivaldi is evidence for the second. It is not evidence for the third.

The advance started with a defensive problem

Lyman is a rail junction on the northern approach to the Sloviansk–Kramatorsk agglomeration, the last large Ukrainian-held urban area in Donetsk oblast. Russia took it in May 2022 and lost it in the counteroffensive that October. The ground north and east of it — forest, river valleys, small villages — has been fought over ever since.

By 2025 the problem there was erosion rather than collapse. Russian forces had established positions west of the Chornyi Zherebets, and small infantry groups were infiltrating past Ukrainian positions into the rear. The commander of the 66th Mechanized Brigade, which has held the Lyman direction since 2022, told Ukrainska Pravda in October 2026 that his brigade lost Serednie and Shandryholove that autumn and winter. He blamed weak company leadership and fragmented command: in summer 2024 his sector of about 38 kilometres was held by five of his own battalions and some fifteen attached ones, many under-trained or under-strength. Ukrainian reporting described a Russian salient west of the river, growing for two years and overlooking the approaches to Sviatohirsk. Russia’s General Staff said on 4 July 2026 that the capture of Lyman — which it calls Krasny Liman — was being completed. It was not; Lyman stayed in Ukrainian hands throughout.

The corps’ own account of its starting point is sharper and should be read as a claim. Its deputy chief of staff said in June 2026 that when the corps began deploying into the Borova–Lyman zone, on 4 June 2025, Russia was taking about 70 square kilometres a month there, with infiltration up to ten kilometres deep, and that both stopped under its command. No independent series for the sector before 2026 was found to check the figure. The problem the operation had to solve is nonetheless clear from both sides’ accounts: before ground could be retaken, the sector had to stop losing it.

First, stabilise the front

The corps describes its first year as stabilisation, and most of what it says it did is about command rather than combat. Tactical decisions were pushed down to brigade and battalion commanders — “we stopped meddling in the brigades’ affairs”, Biletskyi told Hromadske — while the corps took the operational tasks: planning, the deep strikes, the reserves. Reserves were no longer spent plugging tactical gaps a few kilometres behind the line. The corps introduced a battle-captain role for real-time control and, in the 66th Brigade’s account, sent its own officers to mentor brigade staffs; the 66th began running its own basic training in early 2026. A joint headquarters coordinated special-operations and reconnaissance units against infiltration under one plan.

Stabilisation and the operation were not cleanly separate stages. The first announced phase of Vivaldi was itself mostly clearance: on 15 September the corps said 75 square kilometres had been cleared of infiltration and about ten retaken from occupation, with Serednie the first village named. A sector in which infiltration was still being cleared in the first phase of an offensive had not finished stabilising before it began.

Build the system before using it

The question this raises is whether the corps reform of 2025 — Ukraine’s move from brigades fighting under improvised groupings to standing army corps — is what made Vivaldi possible. The evidence cannot settle that. What it does show is the first public case in which a corps can be observed running one plan across several brigades and units it does not own.

The 3rd Army Corps was built in 2025 around the 3rd Assault Brigade, with four further brigades reported to have joined it that summer. In Vivaldi the corps named its own 3rd Assault and 60th Mechanized Brigades, a 125th brigade and the KRAKEN unit, and listed as attached the 66th Mechanized Brigade, special operations forces, Defence Intelligence’s Artan unit and the Border Guard. Those words matter, and Vigil’s records keep them apart. The 66th’s commander calls it a “non-native” brigade subordinate to the corps; the others are published only as attached. The Force Atlas records the 66th’s relationship as a command arrangement whose level is not stated, and does not draw it, or any other attached unit, as a member of the corps.

The corps also describes a surveillance and strike system built around drones: reconnaissance feeds shared in a single command room, separate planning and execution groups, mixed reconnaissance, strike and counter-drone teams, and — on its own figures — ground robots carrying most of the forward supply and evacuation. Artillery appears in these accounts only in passing. That does not mean artillery was absent; it means no source describes what it did. The Kyiv Independent’s reporting from the sector mentions rocket artillery alongside drones. No source read describes an artillery concentration or an ammunition build-up for the operation.

Conceal, then strike before moving

Two steps in the corps’ sequence are better evidenced than the rest, and both came before the ground moved.

The first is concealment. For about four months the operation was not announced, and the corps says its gains were kept out of Ukraine’s own reporting systems; no visible offensive grouping was formed, and a new office ran deception. The method is the corps’ account. The effect is documented by others. Ground Ukraine retook from May appeared on the Institute for the Study of War’s map only in September; DeepState recalculated three months of its figures when the gains were disclosed; and Russian military bloggers complained through the summer that commanders were sending “beautiful reports” up the chain while positions were lost. An operation that the independent maps could not see could not be seen by the other side’s planners either — which is the point of it.

The second is striking first. The corps says it hit Russian logistics, command posts and communications 40 to 150 kilometres deep, pushed Russian unloading points back 40 to 50 kilometres so that infantry had to walk to the front under drone attack, degraded fuel supply into occupied Luhansk, and suppressed Russian drones with electronic warfare before committing armour. The figures it gives — tens of thousands of strikes, some twenty thousand drones neutralised — are its own. The independent line is narrower. ISW assessed on 28 September that Ukraine “likely” neutralised Russian drone defences north of Lyman before sending armour into the third phase, after an intermediate-range strike campaign against logistics and electronic warfare.

Was the contested area isolated? Not in any sense the evidence supports. One village was: Shandryholove, where the 66th’s commander says the Russian garrison was cut off from food and ammunition for weeks and surrendered. At the scale of the sector, Russia kept moving reinforcements in throughout. The accurate word is interdiction — the Russian force’s supply, drones and command were degraded before and during the advance — and it is the step that explains why armour could move at all on a front where, for two years, vehicles near the line have not survived.

Then the map moved

The ground moved in four announced phases. On 15 September the corps disclosed the first, with Serednie retaken. On 20 September it announced Shandryholove and Derylove; on 28 September Nove, Ridkodub and Katerynivka; on 14 October Zelena Dolyna, Lypove and Kolodiazi. Announcement dates are not event dates. Open-source mappers had Nove largely retaken by early June, and the corps’ “phases” were partly a schedule of disclosure.

The totals have to be kept apart, not averaged. The corps’ running count was 85 square kilometres on 15 September, 125 on 20 September, 176 on 28 September and 277 on 14 October; the 125 explicitly included the 75 cleared of infiltration, and on 19 October Biletskyi said infiltration zones were not counted, which leaves the composition of the final figure unclear. The President gave three figures between 30 September and 3 October. The independent figures are lower and count something else. ISW found at least 147 square kilometres around Lyman newly visible in its September mapping; other open-source mappers put the area at between about 150 and over 200 square kilometres by early September. DeepState confirmed the named villages phase by phase and logged 51 square kilometres for the third phase alone.

The month-by-month picture is where the operation most easily misleads. ISW’s assessment of 1 October found Russian-held territory had shrunk by a net 80.81 square kilometres in September — the first month of net Ukrainian gain on its map that Vigil has recorded. But ISW itself said much of that was ground retaken over preceding months becoming visible. DeepState, recalculating for the same late disclosures, put September at a Russian net gain of 154 square kilometres. The two maps disagree on the month’s direction, and both are right about what they measure.

What makes that operational?

Ground alone does not make movement operational. What would is an effect on the opposing force beyond the ground it lost: an objective it must abandon, reserves it must move, an adjacent sector where pressure changes, or movement that continues past its first objective and holds. The figure below takes six such effects and stops each one where its evidence stops.

One effect clears the bar on independent assessment. Russia’s 2026 plan, as ISW and Ukrainian commanders describe it, was a pincer on the Sloviansk–Kramatorsk agglomeration: a northern arm from the Lyman direction and a south-western arm from Dobropillia, meeting near Barvinkove. On 9 October ISW assessed that Vivaldi had dismantled the northern arm, forced Russia to abandon plans to advance on Sloviansk from the north-east, and amounted to an operational failure for Russia. Russian commentators conceded as much in their own terms: the military blogger Rybar wrote that “Sloviansk is postponed”. That is an operational effect, at the scale of one approach to one objective, and Vigil’s standing question on the eastern front now records it as an observable met.

The rest is thinner. The corps says Russia moved part of the 1st Guards Tank Army from the Kupiansk-Vuzlovyi direction to meet the operation, easing pressure there; no Russian or independent source read confirms the move, and ISW said only that units elsewhere “may” have been redeployed. No adjacent sector’s change is independently connected to the operation. And whether the ground holds is not yet observable: the operation ended a week before this Note, and ISW called the situation “fluid”.

The Russian response

Russia’s official voice said almost nothing. No defence-ministry statement on the operation was found. The state’s claims ran the other way: President Putin said on 1 September that Sviatohirsk had been taken, a claim the corps refuted days later with video from the town centre, and the General Staff had claimed Lyman in July. Russian military bloggers were franker. Through the summer some complained of withdrawals and false reports; in September Rybar warned of “problems near Lyman”; by October several conceded that the approach to Sloviansk had been set back, while arguing that the war was not changed by it.

What Russia did with its forces is the more telling evidence, and it cuts against the larger reading of Vivaldi. According to Meduza, Russia sent its 76th Air Assault and 90th Tank Divisions not to Lyman but to the Dobropillia sector, and its 3rd Combined Arms Army kept pushing toward Sloviansk from the east largely on its own. Biletskyi himself said on 19 October that Dobropillia is now Russia’s main axis. Russia did not stop trying to reach the agglomeration; it stopped trying from one direction.

Did the eastern front stop being attritional?

No. Vigil’s standing assessment of the eastern front, set in September, was “attritional”: slow Russian advances at high cost, no marker town changing hands, no new axis. After Vivaldi that assessment stands, with one local exception that it now states. North of Lyman a Ukrainian operation produced movement with an effect beyond its ground. Everywhere else the pattern held: Russia’s main effort continued from Dobropillia, Meduza found the Ukrainian advances no deeper than about ten kilometres in any sector, no marker town changed hands, and the two independent maps disagree on whether September was even a month of net Ukrainian gain.

The distinction is the finding. A front can carry a local episode of manoeuvre without ceasing to be attritional, and Vivaldi is that episode. What would change the assessment is set out on the Watch question: a second month of net Ukrainian gain on independent mapping, a marker town changing hands, or effects like Vivaldi’s assessed on a second axis or outlasting the winter. ISW’s October figure, due in early November, is the first test.

What Vivaldi does not prove

It does not prove that Ukraine has restored manoeuvre warfare. It shows that one corps, in one sector, could create the conditions for movement by first taking away the other side’s eyes and supply. It does not prove that the corps reform caused the success: the reform made a corps-level plan possible, and Vivaldi is the first public case of one, but a single case cannot separate the structure from the people, the resources and an exposed Russian salient. It does not prove that drones replaced artillery, which is simply under-described. It does not prove that every claimed square kilometre is real, and it does not prove that Russia’s offensive capacity has collapsed — Russia’s effort at Dobropillia continued throughout.

Nor does it prove that the method travels. Christopher Tuck of King’s College London has noted that the 3rd Corps’ techniques have not always transferred to other units, and that the corps is unusually skilled at telling its own story. One successful operation changes one sector and one Russian plan. Whether it changes more depends on whether it can be done again, somewhere else, by someone else — and that has not yet been observed.