Guide
Is Solana's 65,000 TPS Real? Theoretical Max vs. Actual Throughput
Bottom line: No — 65,000 TPS is a lab benchmark, not what mainnet runs today
“Solana does 65,000 transactions per second” is one of the most repeated numbers in crypto — and it's misleading if read as a live speed limit. 65,000 TPS is a benchmark figure for simple, optimized transfers under test conditions, first popularized in Solana's own engineering write-ups. Real mainnet throughput, as tracked by independent explorers in mid-2026, typically runs at roughly 1,500–2,000 transactions per second in normal operation, with short bursts recorded up to around 6,000+ TPS — nowhere near 65,000. That gap isn't a cover-up; it reflects how the benchmark was measured versus how a live, permissionless network actually behaves under real demand.
Key takeaways
65,000 TPS is a benchmark, not a live ceiling. It comes from Solana's own reporting on simple SPL transfers under optimized test conditions — not a number the public network sustains day to day. Real mainnet TPS in mid-2026 runs roughly 1,500–2,000 typically, with recorded peaks around 6,000+ in short windows (source: Chainspect, live tracker). Actual TPS measures demand, not capacity — Solana itself has said mainnet usually runs below its available capacity. Firedancer, a new validator client live on mainnet since December 2025, has demonstrated 600,000–1,000,000+ TPS in controlled, multi-node testbeds — a real step toward higher real-world throughput, but not an instant jump to six figures on the public network.
Where the "65,000 TPS" number actually comes from
The figure traces back to Solana's own performance benchmarking on simple transactions — basic token transfers — run under optimized conditions, not the mixed, complex transaction load a live network processes every day. Solana's official Network Performance Report is explicit that the headline number is a demand-vs-capacity distinction: actual network TPS "is not a reflection of network capacity, but of demand for transaction throughput," and mainnet has historically operated below its available capacity in nearly all periods. In other words, the network wasn't hiding a slowdown — the 65,000 figure was never a promise about daily mainnet speed to begin with.
It's also worth noting the number has moved around over the years: Solana's original whitepaper projected an even higher theoretical ceiling, and 65,000 became the figure most commonly cited in later benchmarks and press coverage as the Solana TPS number — which is how it hardened into a "myth" of guaranteed real-world speed.
| Metric | Approx. value | What it actually measures |
|---|---|---|
| Often-cited theoretical benchmark | 65,000 TPS | Simple SPL transfers, optimized lab/test conditions — not mainnet demand |
| Typical real-world throughput (mid-2026) | ~1,500–2,000 TPS | Live mainnet activity (vote + non-vote transactions), fluctuates with demand |
| Recorded short-window peak | ~6,000+ TPS | Brief bursts during high-demand periods, not sustained |
| Firedancer synthetic benchmark | 600,000–1,100,000+ TPS | Controlled multi-node testbed (six nodes, four continents) — not public mainnet conditions |
(Live figures fluctuate constantly and change by the hour — treat the mid-2026 numbers above as a snapshot, not a fixed constant.)
Why the real number is so much lower than the benchmark
A few things explain the gap, and none of them mean Solana is "slow" in absolute terms — it's still far faster than most general-purpose blockchains for everyday transfers. See why Solana is fast and Proof of History for how the architecture is designed to move quickly in the first place. The gap between the benchmark and daily reality comes down to:
- Demand, not a hard ceiling. As Solana's own reporting puts it, live TPS reflects how much block space is actually being used — mainnet usually has room to spare rather than running maxed out.
- Vote vs. non-vote transactions. Every validator continuously submits "vote" transactions to participate in consensus. Raw TPS counts can include this consensus overhead alongside genuine user activity (transfers, swaps, NFT mints), which is why trackers increasingly report non-vote TPS separately as the more meaningful measure of actual usage.
- Lab conditions vs. a real, permissionless network. A benchmark run on a small number of high-spec, co-located machines processing simple transfers isn't the same environment as hundreds of independently operated validators worldwide, running varied hardware, processing a mix of simple and complex transactions, and propagating blocks across real-world network latency.
- Historical congestion episodes. During past periods of extreme demand (meme-coin trading spikes, NFT mint rushes), the network has seen degraded performance and, at times, outages — see Solana's outage history for the details and what's been done since.
A faster benchmark doesn't move mainnet overnight
A validator client hitting huge numbers in a controlled test is a real engineering milestone, but it doesn't instantly change what the live, public network sustains. Gains show up gradually as new software is adopted across a majority of independently run validators — not the moment a demo goes well.
Is Firedancer closing the gap?
Firedancer, an independent validator client built by Jump Crypto, is the most concrete real-world effort to shrink this gap. In a widely reported demo at Solana's Breakpoint 2024 conference, Firedancer processed over 1 million transactions per second in a controlled testbed spanning six nodes across four continents — and in earlier hybrid testing ("Frankendancer"), it handled over 600,000 TPS after deduplication. Those are testbed numbers, not mainnet figures.
On the public network, Firedancer began producing mainnet blocks in phases through late 2025, with a full launch reported around December 12, 2025, and adoption has been climbing since — running on roughly a fifth to a quarter of active validators by mid-2026, with further rollout planned alongside other network upgrades. That's meaningfully closing the software bottleneck (client diversity and raw processing speed), but real, sustained mainnet TPS gains arrive as adoption spreads network-wide, not as a single switch-flip to six-figure throughput.
What to read next
- Basics → What Is Solana · What Is SOL
- How it's built to be fast → Why Is Solana So Fast · What Is Proof of History · What Is Firedancer
- More myth-checks → Does Solana Have a Halving? · Is Solana Decentralized?
- Reliability → Solana's Outage History
Frequently Asked Questions
Q. Is Solana's 65,000 TPS real? A. Not as a live speed limit. It's a benchmark for simple transfers under optimized test conditions. Real mainnet throughput in mid-2026 typically runs roughly 1,500–2,000 TPS, with short peaks around 6,000+.
Q. Why is real Solana TPS so much lower than 65,000? A. Live TPS reflects actual demand for block space rather than a hard capacity ceiling, historically includes consensus "vote" transactions alongside user activity, and runs on hundreds of real-world validators — a very different environment from a controlled benchmark.
Q. Will Firedancer get Solana to 65,000+ TPS on mainnet? A. Firedancer has demonstrated over 1 million TPS in controlled testbeds and has been live on mainnet since December 2025 on a growing share of validators. It's a genuine step toward higher sustained throughput, but network-wide gains roll out gradually as adoption spreads — not instantly.
Q. What's the difference between "TPS" and "non-vote TPS"? A. Raw TPS can include validators' consensus "vote" transactions. Non-vote TPS isolates real user activity — transfers, swaps, NFT mints — and is the more meaningful gauge of actual network usage.
Sources
- Solana Foundation — Network Performance Report: July 2023: https://solana.com/news/network-performance-report-july-2023
- Chainspect — Solana TPS, Max TPS, Block Time & Time to Finality (live tracker): https://chainspect.app/chain/solana
- The Block — Jump Crypto's Firedancer hits Solana mainnet as the network aims to unlock 1 million TPS: https://www.theblock.co/post/382411/jump-cryptos-firedancer-hits-solana-mainnet-as-the-network-aims-to-unlock-1-million-tps
- Solana Compass — Breakpoint 2024 Keynote: Fast Forward from Frankendancer to Firedancer (Kevin Bowers): https://solanacompass.com/learn/breakpoint-24/breakpoint-2024-keynote-fast-forward-from-frankendancer-to-firedancer-kevin-bowers
Important notice
This article is for information only and is not investment advice. Crypto assets (including SOL) carry risks such as price volatility, hacking and network outages. Make decisions at your own responsibility and only with money you can afford to lose.
Sources
- Solana Foundation — Network Performance Report: July 2023
- Chainspect — Solana TPS, Max TPS, Block Time & Time to Finality (live tracker)
- The Block — Jump Crypto's Firedancer hits Solana mainnet as the network aims to unlock 1 million TPS
- Solana Compass — Breakpoint 2024 Keynote: Fast Forward from Frankendancer to Firedancer (Kevin Bowers)
FAQ
- Is Solana's 65,000 TPS real?
- Not as a live speed limit. It's a benchmark for simple transfers under optimized test conditions. Real mainnet throughput in mid-2026 typically runs roughly 1,500–2,000 TPS, with short peaks around 6,000+.
- Why is real Solana TPS so much lower than 65,000?
- Live TPS reflects actual demand for block space rather than a hard capacity ceiling, historically includes consensus "vote" transactions alongside user activity, and runs on hundreds of real-world validators — a very different environment from a controlled benchmark.
- Will Firedancer get Solana to 65,000+ TPS on mainnet?
- Firedancer has demonstrated over 1 million TPS in controlled testbeds and has been live on mainnet since December 2025 on a growing share of validators. It's a genuine step toward higher sustained throughput, but network-wide gains roll out gradually as adoption spreads — not instantly.
- What's the difference between "TPS" and "non-vote TPS"?
- Raw TPS can include validators' consensus "vote" transactions. Non-vote TPS isolates real user activity — transfers, swaps, NFT mints — and is the more meaningful gauge of actual network usage.
本文仅供参考，不构成投资、金融或交易建议。价格为参考值，可能已过时。投资决策请自行判断。
