DIEM, MOR, and JTVO are only worth it if you like risk

July 13, 2026

Not financial advice.

Baseline: ChatGPT plans

• Plus: $0.66/day → ~$26/day usage: ~39x multiplier
• Pro 5x: $3.33/day → ~$130/day usage: ~39x multiplier
• Pro 20x: $6.66/day → ~$520/day usage: ~78x multiplier

Reward without risk: Surplus Intelligence

Just pay for tokens at a 1.4x-200x discounted rate (when calculating yield we'll assume 2x) versus official pricing. Multiplier varies based on suppliers, which vary based on model and time of day.

As of writing, Surplus Intelligence didn't reply to a request for comment as to whether caching discounts are supported.

What about open weights?

When using raw APIs, you usually get 10x more usage with open weight than closed weight models. By either using subscriptions directly (Ollama Cloud, Z.ai) or someone else's subscriptions via Surplus Intelligence, you can get another 10x on top of that.

100x isn't as good as it sounds - you can also find GPT 5.5 with a clean 100x multiplier on Surplus Intelligence. Open weight models are only cheaper when using official APIs.

Introducing staking: DIEM

Normally currency is exchanged for goods and services. In crypto, currency can simply be staked, letting you get it back at will. And if you stake a DIEM, you net $1 of AI daily at Venice rates, or ~$0.5 daily from reselling on Surplus Intelligence, or 365 days * $0.5/day / $1300 = ~14% yield, or doubling its value (remember you can unstake it when done!) 7 years out.

The efficient market hypothesis teaches us that if something is too good to be true, it likely is. There's a big risk here stopping DIEM from doubling its price* and bringing its yield down to normal levels. And I think I know what it is: if Venice shuts down, your token is worth absolutely nothing. It won't give you any more inference, and for that same reason, nobody will want it.

*price would also increase if people unminted DIEM, bringing DIEM supply down and netting them a personal profit

It's MORbin Time

There's this guy Ryan Condron. He co-founded Titan, built Lumerin (general purpose blockchain-powered matchmaking) and Morpheus Network (MOR) there, then stepped down sometime after that. He's now running MORDIEM, an inference market powered by MOR, Venice (DIEM), and Lumerin. MORDIEM has a lot of MOR in custody:

• $78,404 is in 0xd01c1B0eeDBE341c409369177478f2Eabbeee848. MORDIEM uses it for user transactions.
• $127,059 is in 0xC599884ACB421BD548D6fAe444CD075DFB012FBF. MORDIEM uses it for Lumerin transactions.
• $352,330 is staked in 0x42BB446eAE6dca7723a9eBdb81EA88aFe77eF4B9. This contract is called the builders pool.

But that doesn't actually decide how much daily inference you get. Instead, your share of "Total MOR Supply" (all MOR minted ever except for the purpose of funding compute providers) staked is what determines your share of the daily allocation (1%) of the "Compute Balance" (all MOR minted for the purpose of funding compute providers ever).

As of writing, the math comes out so if you stake 190 MOR to MORDIEM, you net $1 of AI daily at Venice rates = ~$0.5 daily from reselling on Surplus Intelligence AND 20% APY. This makes 365 days * $0.5/day / $418 + 0.2 = ~63% yield, or doubling its value in 19 months.

I think here the market is pricing in that betting on MORDIEM is betting on these risks:

• MORDIEM will keep providing MOR suppliers with inference and interest (including the continued existence of the builders pool)
• MORDIEM will keep providing MOR suppliers with the option to take back their MOR (their TOS calls staking "gifting") and MOR holds price
• DIEM holders will keep providing inference to MORDIEM (including Venice holding up inference + MOR holding up MOR payments)

The only way this wouldn't end badly is if MORDIEM gracefully shuts down and lets you take your MOR - since MOR would still be worth something you would've lost nothing.

Jul 15 2026 addendum: JaTeVO

Jatevo is similar to the other providers in that it's crypto, but it's different in that it gives you a daily requests allocation, and that it's a reseller of ChatGPT subscriptions. Single requests to Sol can cost $14.98 or $4.56, so let's look at historical Surplus Intelligence charges per request instead.

So yield is ($0.004/day * 365 days) / $1 = 146%.

Remember GitHub Copilot's monthly premium requests? Or CrofAI's daily request plans? Per request billing isn't sustainable. And Jatevo is a vibe coded website leaking too many implementation details, with no $65 mil series A or rich founder to bear the load. I'm bearish.

Conclusion

If you pay as you go:

• $20/mo or $100/mo GPT → ~39x multiplier
• $200/mo GPT → ~78x multiplier
• Surplus Intelligence → up to 100x multiplier

If you invest in crypto, and 1 year later things go well:

• $1300 DIEM, AI use → got $ from DIEM appreciating AND got $365 in AI (would cost $260-$1.83 on Surplus Intelligence)
• $1300 DIEM, reselling → got $ from DIEM appreciating AND got $182.5
• $418 MOR, AI use → got $ from MOR appreciating AND got $85 from MOR interest AND got $365 in AI (would cost $260-$1.83 on Surplus Intelligence)
• $418 MOR, reselling → got $ from MOR appreciating AND got $85 from MOR interest AND got $182.5
• $100 JTVO, AI use → got $ from JTVO appreciating AND got 36.5k GPT requests (could cost anything on Surplus Intelligence, realistically between $3.65 and $365)
• $100 JTVO, reselling → got $ from JTVO appreciating AND got $146

If you invest in crypto, and 1 year later things go awry:

• $1300 DIEM → Venice shuts down, you only ever made $182.5, you lost $1117.5 (86%)
• $418 MOR → MORDIEM shuts down hard, you only ever made $182.5, you lost $236 (56%)
• $100 JTVO → Jatevo shuts down, you made $146