Quality vs bitrate from independent tests in 2026:
| Codec | File size at same VMAF | Encoding speed | Decoding |
|---|
| H.264 (libx264) | Reference | Fast | Universal |
| HEVC (libx265) | 50% | Slow | Modern devices |
| AV1 (SVT-AV1) | 35% | Reasonable | Modern devices |
| VVC (VVenC) | 30-35% | Slow | Few devices |
| AVC (libx264) at same bitrate | Worse | n/a | n/a |
VVC and AV1 are roughly comparable on file size at the same quality. The differences are in 5-8% range, which doesn't decide the question. Other factors do.
| Device class | H.264 | HEVC | AV1 | VVC |
|---|
| iPhone 8-13 | Yes | Yes | No | No |
| iPhone 14 | Yes | Yes | No | No |
| iPhone 15 / 16 | Yes | Yes | Yes | No |
| Pixel 6-7 | Yes | Yes | Limited | No |
| Pixel 8-9 | Yes | Yes | Yes | No |
| Galaxy S22-S24 | Yes | Yes | Yes | Limited |
| Galaxy S25 | Yes | Yes | Yes | Yes (some markets) |
| RTX 30 series | Yes | Yes | Decode only | No |
| RTX 40 series | Yes | Yes | Yes (encode + decode) | Decode only |
| Intel ARC | Yes | Yes | Yes (encode + decode) | Decode only |
| Apple M1 | Yes | Yes | Decode only | No |
| Apple M3 / M4 | Yes | Yes | Decode + encode | No |
| Smart TVs (2024+) | Yes | Yes | Roku/Google: Yes | LG OLED: Yes |
| Browsers (Chrome 100+) | Yes | Yes | Yes | Behind flag |
AV1 is winning the hardware ubiquity race. Apple's iPhone 15 commitment was the inflection point. By 2027, AV1 will be on most consumer devices. VVC is getting hardware decode in Samsung 2025 flagships and select smart TVs, but the rollout is slower.
This is where VVC stumbles.
AV1 patent license: AOMedia Patent License. Royalty-free. ~30 patent-holding companies cross-license to one another and to the world. Includes Google, Cisco, Mozilla, Intel, Microsoft, Apple, Netflix, Samsung, AMD.
VVC patent license: three competing pools. MPEG LA charges $0.32/unit on devices, $0.0024/sub-month on content. Access Advance charges similar. Velos has separate terms. Total cost is probably $0.50-1.00 per unit shipped, plus content licensing.
For a streaming service with 10 million subscribers shipping VVC for a year: roughly $290,000 in content licensing alone. AV1 cost: $0.
This is the same trap that killed HEVC's growth as a delivery codec. Streaming services moved to AV1 specifically to avoid HEVC's licensing.
| Codec | Encoders (open source) | Production-ready? |
|---|
| AV1 | SVT-AV1, libaom, rav1e | Yes (since 2022) |
| AV1 | Hardware (NVENC, QSV, AMF) | Yes (since 2024) |
| VVC | VVenC (Fraunhofer, BSD) | Improving (2025+) |
| VVC | x266 (Open source) | Early stage |
| VVC | Hardware encoders | Few, expensive |
For AV1 production: SVT-AV1 is the go-to. Hardware encoders on RTX 40 / ARC / RDNA3 GPUs make AV1 batch work fast.
For VVC production in 2026: VVenC is the only practical open-source encoder. Encoding is roughly 10x slower than libaom-av1 at similar quality. No widely-available hardware encoders yet.
For background on AV1 encoder choice, see AV1 Encoding in 2026.
The actual betting is happening at scale:
AV1 deployments (2026):
- Netflix (encoded library at ~30% AV1, expanding)
- YouTube (AV1 default for most accounts)
- Twitch (testing AV1 ingest from select Partner streamers)
- Vimeo (AV1 in Pro tier delivery)
- Disney+ (AV1 4K HDR for some content)
VVC deployments (2026):
- Limited 8K broadcast experiments (Japan NHK)
- DVB-C and DVB-S2 European broadcasters experimenting
- No major OTT streaming platform yet
The streaming industry is voting with deployments. AV1 is the practical answer in 2026. VVC may catch up in broadcast, where its specific advantages (8K, immersive video coding extensions) matter.
VVC's specific advantages over AV1:
8K and beyond: VVC was designed with 8K as the primary target. It scales better than AV1 for very-high-resolution content. For 8K streaming or broadcast, VVC compresses noticeably better.
Immersive video: VVC has dedicated extensions for VR/AR (multi-view, depth maps). AV1 added similar support later via separate specifications.
Lossless mode: VVC's lossless mode is more efficient than AV1's lossless. For archival, this matters.
Broadcast standards alignment: VVC fits into existing DVB and ATSC broadcast standards more naturally. The MPEG track has 30 years of broadcast integration; AOMedia track is newer and web-focused.
If your use case is 8K, VR, broadcast, or archival: VVC is worth tracking. If your use case is web streaming, mobile delivery, or general-purpose video: AV1.
AV1's specific advantages over VVC:
Cost: Royalty-free. Save the per-unit and per-subscriber licensing fees.
Browser support: AV1 plays in Chrome, Firefox, Edge, Safari (17+) without flags. VVC is behind flags or unsupported.
Hardware decode ubiquity: iPhone 15+, recent Pixel, recent Galaxy, recent Macs, RTX 40 series. VVC hardware is rare in 2026.
Web ecosystem: Major streaming services adopted AV1. Smart TV manufacturers ship AV1 decoders. Mobile chipsets bake it in.
Encoder maturity: SVT-AV1 is production-grade in 2026. VVenC is improving but slower and less feature-complete.
For most production targets in 2026, AV1 is the right bet.
For specific use cases:
| Use case | 2026 codec | Fallback |
|---|
| YouTube delivery | H.264 + AV1 | HEVC |
| Streaming service delivery | AV1 | HEVC |
| Mobile app delivery | H.264 + AV1 (iOS 17+, Android 14+) | HEVC |
| Web embedding | H.264 + AV1 | HEVC |
| Archival masters | ProRes / DNxHR / FFV1 | n/a |
| 8K delivery | VVC (when ready) | AV1 |
| Live streaming | H.264 (still) | AV1 hardware |
| Email attachments | H.264 | n/a |
H.264 remains the universal floor. AV1 is the upgrade path for everything except live and 8K. VVC is the niche choice for specific high-resolution targets.
For VVC encoding with VVenC:
vvencFFapp \
-i input.yuv \
-s 1920x1080 \
-r 30 \
--preset slow \
--qp 30 \
-o output.266
VVenC takes raw YUV input, so source must be pre-decoded. Combine with FFmpeg:
ffmpeg -i input.mp4 -f yuv4mpegpipe -pix_fmt yuv420p - | \
vvencFFapp -i - -y4m --preset slow --qp 30 -o output.266
For AV1 with SVT-AV1, see AV1 Encoding in 2026.
VVC files won't play: most browsers don't support VVC in 2026. Test with VLC 4+ or play in a dedicated VVC-compatible player like Fraunhofer's reference player.
AV1 vs HEVC quality dispute: at the same bitrate, both produce similar quality on consumer content (talking heads, action). On scientific content (medical imaging, microscopy), HEVC's 12-bit support gives it an edge. AV1 supports 12-bit but encoder support is rough.
Slow encoding: VVenC is roughly 50x slower than libx264. Plan accordingly. For batch work, use cloud GPUs or wait for VVC hardware encoders (expected 2026-2027).
Hardware decode confusion: don't trust marketing claims. Verify on real devices with real test files. Some "AV1 decode" hardware fails on 10-bit or HDR variants.
For a deeper look at AV1 specifically, see AV1 Encoding in 2026.
YouTube re-encodes everything regardless. Upload H.264 at high bitrate for best quality after their re-encode. AV1 source upload is fine but doesn't gain you anything; YouTube doesn't pass through your encoded file.
The streaming industry has effectively chosen AV1. VVC may grow in broadcast and 8K niches. Web video infrastructure is overwhelmingly AV1-bound. Catching up requires a Netflix-or-YouTube-scale endorsement that hasn't happened.
AV2 (next from AOMedia) is in development, targeting 2026-2027. EVC and LCEVC are alternative codecs that gained limited traction. For practical 2026 work, the AV1 vs VVC choice is the question.
Yes. Use the <source> element with multiple types in HTML5 video. Browsers pick the first format they support. Production CDNs (Cloudflare, Akamai, Fastly) handle on-the-fly format negotiation.
VLC 4.0+ has VVC support. Download a VVC sample from Fraunhofer's website. Play locally. Browsers don't render VVC reliably yet.
Hardware encoding with RTX 40 / ARC / RDNA3 makes 1080p60 AV1 live possible. Twitch is testing it with select partners. For most live setups, H.264 still wins on broad device compatibility. AV1 live is the future, not the present.
For 2026 web video and most consumer delivery: AV1. Royalty-free, broad hardware support, mature encoder. For 8K broadcast, immersive video, or scenarios where MPEG ecosystem alignment matters: VVC, accepting the licensing cost. H.264 remains the universal floor. Our video compressor handles current-generation delivery; AV1 is the upgrade most workflows should plan around.
