Cosy Octopus vs Octopus Agile: which is better for a heat pump in 2026?
Both tariffs are from Octopus Energy and both can deliver cheap electricity for heat pump owners. But they work completely differently. Cosy gives you the same three cheap windows every day, predictably. Agile gives you 48 half-hourly prices that change daily — sometimes very cheap, sometimes expensive. The gap between the two can be hundreds of pounds a year depending on how much automation you're willing to set up.
Two tariffs, one supplier, very different approaches
Octopus offers both Cosy Octopus and Agile Octopus. They're not really competing products — they're designed for different types of household. But heat pump owners regularly ask which one is better, so it's worth a direct comparison.
| Cosy Octopus | Octopus Agile | |
|---|---|---|
| Rate structure | 3 fixed bands daily | 48 half-hourly slots, changes every day |
| Cheapest rate | ~14.5p/kWh (off-peak windows) | Can go below 5p/kWh or even negative |
| Most expensive rate | ~51.7p/kWh (16:00–19:00 peak) | Capped at ~100p/kWh in extreme peaks |
| Predictability | High — same windows every day | Low — rates change daily |
| Scheduling effort | Moderate — align to 3 fixed windows | High — needs automation to exploit cheapest slots |
| Exit fee | £25 (6-month fix) | None |
| Best suited to | Heat pump households with smart scheduling | Tech-savvy households with automation or battery |
How Cosy Octopus works
Cosy Octopus is designed specifically for heat pump homes. Three fixed cheap windows run every day — 04:00–07:00, 13:00–16:00, and 22:00–00:00 — at approximately 51% below the standard day rate (~14.5p/kWh at April 2026 rates). Outside those windows you pay the standard cap rate. Between 16:00–19:00 you pay a peak rate roughly 50% above the day rate (~51.7p/kWh).
The windows never change. The rates move quarterly when the Ofgem cap changes, but the structure — which hours are cheap, which are peak — is fixed. That makes it straightforward to set your heat pump schedule once and leave it.
How Octopus Agile works
Agile prices electricity in 48 half-hourly slots, updated daily from wholesale market data. Tomorrow's prices are published at around 4pm today, visible in the Octopus app. On a typical day, overnight and early morning slots might run at 8–15p/kWh; the late afternoon 16:00–19:00 peak period might run at 30–60p/kWh or higher in cold, low-wind weather.
On genuinely good days — high wind, mild temperatures, excess renewable generation — Agile prices can go negative, meaning Octopus pays you to consume electricity. These negative-price events happen a few dozen times per year and are valuable if you can exploit them automatically.
The 100p/kWh cap protects against the most extreme spikes, but prices above 40p/kWh are uncommon rather than rare, particularly in winter.
The heat pump question: which is actually cheaper?
The honest answer is: it depends on how much automation you have.
Without automation (manual scheduling only):
Cosy wins. With a fixed schedule set in your heat pump controller, you capture the Cosy off-peak rate reliably. On Agile without automation, you'd need to check tomorrow's prices daily and adjust your schedule — impractical for most people. The average Agile rate without any load-shifting is close to the standard cap or above it.
With basic Home Assistant automation:
Agile starts to compete. A simple automation that runs your hot water cylinder during the cheapest overnight Agile slots — identifiable from the daily price feed — can capture rates significantly below Cosy's off-peak. The Octopus Energy integration for Home Assistant makes this relatively straightforward to set up.
With full Home Assistant automation and a buffer cylinder:
Agile typically wins. A setup that automatically triggers hot water and space heating preheat during the cheapest 3–4 slots of the day, avoids slots above a set threshold, and responds to negative-price events can achieve an effective blended rate of 12–16p/kWh across the year — competitive with or better than Cosy's off-peak rate, without paying Cosy's peak rate at all during high-price Agile events (because the automation avoids them).
The peak rate comparison
This is where the tariffs diverge most sharply.
Cosy peak (16:00–19:00): ~51.7p/kWh, every single day, same time, guaranteed. If your heat pump or immersion heater runs during this window, you pay 51.7p for every kWh.
Agile peak: prices during 16:00–19:00 vary. On a mild summer day, the peak slots might only reach 20–25p/kWh — cheaper than Cosy's peak. On a cold January evening with low wind, they might reach 60–80p/kWh — more expensive than Cosy. The average across the year is probably 30–40p/kWh during that window.
With automation, you can pause your heat pump when Agile prices exceed a threshold. Without automation, you're exposed to whatever the market delivers during that window.
Who should consider Agile
You have Home Assistant or equivalent. The Octopus Energy integration is mature, well-documented, and has active community support. If you're already using Home Assistant for your heat pump, adding Agile price-aware automations is a natural extension.
You have a large buffer cylinder or thermal store. Agile's cheapest slots tend to cluster in the early hours of the morning (00:00–06:00) and sometimes during high-renewable midday periods. A large cylinder lets you load up on cheap energy during those windows and coast through expensive periods. The longer the thermal store's hold time, the more Agile rewards you.
You're on Octopus already and don't have a long commitment. Agile has no exit fee; Cosy now has a £25 fee and a 6-month commitment. If you want to experiment with wholesale pricing before committing, Agile lets you test it without the friction of Cosy's fixed term.
You find the Agile community useful. The Octopus Agile user community is active on Reddit and the Octopus forum. Tools like Agile Insights and the energy-stats.uk Agile dashboard give you daily context on whether today is a good or bad Agile day. That community support lowers the barrier to getting value from the tariff.
Who should stick with Cosy
You want set-and-forget simplicity. Set your hot water and space heating schedule once to the Cosy windows, leave it, and capture consistent savings. No daily price checking, no automation debugging, no risk of an unexpected price spike.
Your heat pump controller supports time-based scheduling but not API integration. Many modern heat pumps (Vaillant, Daikin, Samsung) have schedulable controllers that can be set to run during fixed windows. That works perfectly for Cosy. Exploiting Agile's dynamic pricing with those controllers requires a separate automation layer.
You have no interest in home automation. That's a completely reasonable position. Cosy delivers good savings — £200–£400/year on an actively managed setup — without requiring any technical setup beyond a heat pump schedule. Agile's ceiling is higher but its floor without automation is lower.
The honest verdict
Cosy is the better default choice for heat pump owners. It's purpose-built for the use case, the cheap windows are long enough to cover hot water and space heating preheat, and the savings are predictable.
Agile has a higher ceiling — a well-automated household can do materially better than Cosy's off-peak rate, and there's no six-month commitment or exit fee. But realising that ceiling requires Home Assistant or equivalent automation, a large enough thermal store to exploit overnight cheap slots, and enough comfort with wholesale price volatility to handle the occasional expensive day.
The right question isn't "which is better" — it's "which suits how I want to manage my energy". If you want to optimise and automate, Agile has more upside. If you want reliable savings without ongoing management, Cosy is the stronger choice.
Both are from Octopus, so you don't need to switch supplier to try either one. And you can compare both against your actual usage data before committing to either.
See how Cosy Octopus, Agile, and every other heat pump tariff compares against your real half-hourly usage on Heat Pump Tariffs.