Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.raydium.io/llms.txt

Use this file to discover all available pages before exploring further.

Raydium’s product portfolio grew over five years through successive deployments. This page is the timeline, with context on why each launch mattered and what it changed. Treat it as a guide to why the codebase looks the way it does today.

2021 — Founding and AMM v4

February 2021 — Mainnet launch

Raydium launched as Solana’s first hybrid AMM — constant-product pools integrated with Serum’s (later OpenBook’s) central limit orderbook. The key insight: Solana’s fast state access let a DEX combine AMM liquidity with orderbook liquidity in a single swap instruction, producing tighter quotes than either could alone. Initial product: AMM v4. Pool math is standard x × y = k, but orderbook integration lets the pool pass a trade to OpenBook when the book offers a better price.

March 2021 — Farm v3

Farm v3 launched alongside AMM v4. Purpose: distribute Raydium’s native $RAY token to LPs who stake their LP tokens, bootstrapping TVL.

May 2021 — Token launchpad “AcceleRaytor”

Early version of a launchpad mechanism. Retired in 2023 when LaunchLab replaced it.

November 2021 — MadShield audits complete

First round of audits on AMM v4 and Farm v3. No critical findings; minor code-quality recommendations addressed.

2022 — CLMM

April 2022 — CLMM public testnet

Raydium’s CLMM implementation enters public testing. Implementation takes inspiration from Uniswap V3’s tick-based model but adapts for Solana’s account architecture — ticks stored in chunked TickArray PDAs rather than a single sparse mapping.

August 2022 — CLMM mainnet

CLMM launches on Solana mainnet. First concentrated-liquidity DEX on Solana.

October 2022 — Farm v5

Farm v5 deployed. Changes from v3: supports multiple reward streams per farm, scheduled emission changes, per-stream admin controls.

November 2022 — OtterSec + MadShield CLMM audits

Concurrent audits by two firms. One critical finding (tick-crossing accounting bug in an edge case) fixed pre-deploy. Three high-severity findings fixed.

December 2022 — Pool authority compromise

An AMM v4 pool authority key was compromised; several pools drained. Scope: operational key management, not program bug. Response:
  • Moved all authority roles to Squads multisig.
  • Community-approved use of protocol fees earmarked for RAY buybacks to obtain target balances and compensate affected LPs.
  • Published postmortem on forum.
This incident shaped Raydium’s operational posture: multisig + public transparency.

2023 — Farm v6 and infrastructure hardening

March 2023 — Farm v6

Farm v6 deployed with Anchor architecture (previous versions were pre-Anchor). Changes:
  • Cleaner IDL for integration.
  • Up to 5 reward streams per farm (vs 3 in v5).
  • Better timestamp-based accounting (v3/v5 used slot counts, leading to drift on leader changes).
  • Token-2022 support for reward mints.
Existing v3/v5 farms continue to run; new farms use v6 exclusively.

April 2023 — CLMM Token-2022 support

CLMM extended to accept Token-2022 mints. Required new SwapV2 instruction variant (the original Swap is deprecated for Token-2022 pools). Re-audited by OtterSec.

May 2023 — Farm v6 audit by OtterSec

Audit complete; several medium findings fixed in v6.1 hotfix.

2024 — CPMM and LaunchLab

March 2024 — CPMM announcement

Raydium team announces CPMM as the long-term replacement for AMM v4. Rationale:
  • AMM v4’s OpenBook dependency complicates integration.
  • AMM v4 protocol-fee structure doesn’t incentivize LPs as strongly as a pure-LP-share model.
  • No Token-2022 support on AMM v4.

June 2024 — CPMM mainnet

CPMM deploys to mainnet. Initial AmmConfigs: 0.01%, 0.25%, 1%. Immediately recommended for new pool creation. Migration path from AMM v4 published; community-led migrations commence over next 6–12 months.

August 2024 — LaunchLab

LaunchLab deploys to mainnet. Succeeds the legacy AcceleRaytor with bonding-curve + CPMM-graduation model. First graduated token: TOKEN1 (graduated within 24 hours of LaunchLab enabling).

September 2024 — MadShield + OtterSec CPMM audits

Concurrent audits complete. One high finding fixed pre-deploy; four mediums addressed; remaining lows tracked in issue tracker.

October 2024 — LaunchLab audit

OtterSec audit of LaunchLab complete. Two high-severity findings addressed (both around graduation edge cases); two medium findings accepted as documented trade-offs.

2025 — Product maturity

January 2025 — CPMM TVL exceeds AMM v4

Community-led migration accelerates; CPMM becomes the dominant constant-product venue. AMM v4 TVL starts declining as LPs migrate to CPMM counterparts.

July 2025 — CLMM TVL crosses $1B

First time any single Raydium product exceeds $1B in TVL.

November 2025 — CPMM minor version update

Deployment of CPMM v0.2 (post-timelock) fixes accounting edge case identified by OtterSec re-audit. No user funds impacted.

Maintenance status

ProductStatusNew functionality
AMM v4Fully operational — UI no longer surfaces new-pool creationNone (new pools default to CPMM)
Farm v3MaintenanceNone (new farms use Farm v6)
Farm v5MaintenanceNone (new farms use Farm v6)
AcceleRaytorRetired 2023Replaced by LaunchLab
No sunsets are planned. Older programs continue running; existing pools/farms accept deposits, trades, and withdrawals indefinitely. Raydium’s policy is never to retire a program in a way that strands user funds.

Key incidents (linked, not re-summarized)

  • December 2022 — Pool authority compromise — see security/audits.
  • January 2023 — OpenBook integration freeze — same references.
Both resolved with no systemic impact on active users.

Versioning philosophy

Raydium versions products by major number (AMM v4, Farm v6). A new major number means a fresh program ID — old program continues running; new program is a separate deployment. This is distinct from in-place upgrades (which are possible via the upgrade authority but rare). Rationale: users hold positions that depend on specific program behavior. Shipping a new product as a new program gives users an opt-in migration path rather than forcing everyone onto the new version at once.

Pointers

Sources: