| Edition 01 | 18 August 2026 | Series: Foundations |
It is the study phase — twenty-seven parallel investigations that produce no normative 6G specification whatsoever. Understanding why that is deliberate is the prerequisite for reading everything that follows.
| Quarter | Track | Milestone |
|---|---|---|
| Q2 2026 TSG#112 | SA / CT | Rel-21 timeline agreed; TR 38.914 v1.0.0 approved done |
| Q3 2026 TSG#113 | SA / CT | Rel-20 SA Stage 2 complete |
| Q4 2026 TSG#114 | RAN | Rel-20 RAN1 functional freeze |
| Q1 2027 TSG#115 | RAN | Rel-20 RAN2/3 & RAN4-core functional freeze ⚑ freeze |
| Q1 2027 TSG#115 | SA / CT | Rel-20 Stage 3 freeze (SA/CT) · Rel-21 package + 6G work items approved (Stage 1 freeze) ⚑ freeze |
| Q2 2027 TSG#116 | RAN | Rel-20 RAN4 performance |
| Q2 2027 TSG#116 | SA / CT | Rel-20 ASN.1 & OpenAPI freeze |
| Q3 2028 TSG#121 | RAN | Rel-21 RAN1 functional freeze |
| Q3 2028 TSG#121 | SA / CT | Rel-21 SA Stage 2 complete |
| Q4 2028 TSG#122 | RAN | Rel-21 RAN2/3 & RAN4-core functional freeze ⚑ freeze |
| Q4 2028 TSG#122 | SA / CT | Rel-21 Stage 3 freeze (SA/CT) ⚑ freeze |
| Q1 2029 TSG#123 | SA / CT | Rel-21 ASN.1 & OpenAPI freeze |
Transcribed from 3GPP’s published Release 21 timeline (© 3GPP 2026). Rel-20 is the 6G study phase; Rel-21 is the first 6G normative release.
27 Rel-20 6G studies | 2 Complete so far | Q1 2027 Rel-21 package approval |
3GPP splits a generation into a study phase and a normative
phase. Release 20 is the former. Its 6G output is entirely Technical Reports —
TR 22.870, TR 38.914, TR 38.760-x,
TR 23.801-01, TR 33.801-01 and the rest. Not one of them is
implementable. They exist to converge the industry on what problem is being solved
before anyone writes a bit-level definition.
Roughly half of Rel-20 working-group time goes to these 6G studies. The other half continues 5G-Advanced evolution — which is why Rel-20 reads as two releases stapled together, and why its work plan confuses anyone who misses the split.
Terminology, as agreed across all three TSGs
6GR — 6G Radio6G RAT — the radio access technology itself6G RAN — the radio access network6GC / 6G CN — 6G core network6GS — the 6G system, end to endNot cosmetic. Cross-TSG terminology drift is a standing source of specification defects; fixing it before Stage 2 starts is cheap, fixing it after is not.
| Group | Studies | Anchor deliverables |
|---|---|---|
| SA1 | 1 +1 normative | 22.870 → 22.270 |
| TSG RAN | 1 | 38.914 |
| RAN1–4 | 4 | 38.760-1/-2/-3/-4 |
| SA2 | 1 | 23.801-01 |
| SA3 / SA3-LI | 4 | 33.801-01, 33.703, 33.771, 33.801-06 |
| SA4 | 1 | 26.870 |
| SA5 | 2 | 32.801-01, 32.801-02 |
| SA6 | 2 | 23.801-02, 23.801-03 |
| CT1 / CT3 / CT4 | 9 | 24.860–862, 29.830/832/833, 29.840/841/842 |
| Cross-TSG | 1 | 21.802 |
Two are already done. TR 22.870 — 590+ pages of use cases and service
requirements, rapporteur Xiaonan Shi (China Mobile) — was approved at
TSG#111 in Fukuoka, March 2026. TR 38.914, rapporteur Xiaoran Zhang
(China Mobile), reached v1.0.0 at RAN#112 in Singapore on 14 June 2026. The rest close
between March 2027 and September 2027.
Read the timeline backwards and the logic appears. The ITU-R deadline is fixed. Stage 3 must freeze roughly a year before it, Stage 2 a year before that — which leaves exactly the window Rel-20 occupies, and explains why studies that look leisurely are tightly boxed.
A common misread is that studies are preamble and the normative release is where it counts.
The opposite is closer to true. By the time TR 23.801-01 closes in March 2027, SA2
will have resolved its ~24 key issues into agreed conclusions — and Rel-21 Stage 2 largely
transcribes those into TS 23.xxx procedures. The same holds for the radio: whatever
TR 38.760-1 concludes on waveform, numerology and channel coding becomes the
physical layer.
If you want to influence 6G, the window is now and it closes in early 2027. If you want to predict 6G, the study TRs are the highest-signal documents available.
Rel-20 carries an unusual amount of AI scope, and it is not confined to one group. SA1 wrote
network AI agents, third-party AI agents, and AI model training and inferencing into
TR 22.870 as functional requirements. SA2 has four “beyond connectivity”
work tasks including AI and computing coordination across UE, core and application server. CT3
runs TR 29.832, “Study on the Protocol for Artificial Intelligence in
6G”, rapporteur Apostolos Papageorgiou (Nokia Germany), and is openly
weighing whether to adopt IETF agentic protocols, define 3GPP-specific ones, or assemble
something from Agent2Agent, ANP, AGNTCY and MCP.
Three groups converging on the same question — what is an AI agent to a mobile network? — from requirements, architecture and protocol angles at once. Whether they converge on one answer is the most interesting open question in the release.
Sources
Daily Dose of 6G — tracking 3GPP Release 20's 6G study phase, one study at a time. Timeline after 3GPP's published Release 21 schedule.