Dolphin

Node provider rollout has been going well Our pool of inference nodes running Qwen 3.6 35B have generated over 3.2B tokens so far Total inference bandwidth -> 9400 t/s 28x RTX 4090 12x RTX 5090 8x RTX PRO 6000 & many other cards API access coming soon 🐬

@0xmons Was referring to the base model made by Mistral linked above Still a very popular model despite being a few months old

@dphnAI looks like 14k last month? huggingface.co/dphn/Dolphin-M…

u guys are shilling a ticker for an ai model that is based on an llm from 3 yrs ago????

@0xmons This was built on Mistral small 3.2 vision, the updated version - still has 870k monthly downloads Have a bunch of new models we have been working on that will be releasing soon huggingface.co/mistralai/Mist…

@dphnAI not...that encouraging?

a decentralized inference network seems powerful

@lucaxyzz Guide at dphn.ai/docs RTX 6000 PRO can participate now in running Qwen 3.6 35B Support for smaller GPUs coming soon You can watch the network stats live at datagen.dphn.ai

Interesting to see where $POD and @dphnAI go. How can we get involved on the inference/compute side? We have an RTX PRO 6000 and an RTX 5090 that could potentially be routed into your network @dphnAI

@_0xghost_ @FoFan_eth @AskVenice dm sir, will send you the deck Currently updating website with new docs

@FoFan_eth @dphnAI @AskVenice Great work here! How do you know 100% of rev goes to token burns, is it published anywhere?

$POD - is the @dphnAI's and @AskVenice's bet on Uncensored AI models. I don't know why no one is talking about it. - The General Info Dolphin is an AI Lab, which developing uncensored models & distributed inference. They are already built the most powerfull Uncensored AI Model Yet - Dolphin Mistral 24B Venice Edition. Dolphin Mistral 24B Venice Edition is available as “Venice Uncensored” in the model dropdown and is the new default on the Venice platform. The model is leading by the Censorship Refusal Rate, with only 2.2% of refusals. For example, Grok has 26.67%, Claude Sonnet 3.7: 71.11%, GPT-4o-mini: 64.44%. - @AskVenice & @opentensor & @TargonCompute To develop the model Dolphin brought together Venice and Subnet 4 @TargonCompute for compute power, the model was trained on the subnet. Venice played a role as co-fine-tuning partner, platform integrator, and release coordinator. Finally, we see a development of real unique model fully via DeAI. CryptoAI narrative fits the best here. - Founder and @a16z grant Dolphin, founded by Eric Hartford (@QuixiAI). He started working on Uncensored models since 2023 and launched his first model "dolphin-llama2-7b" in 2023. It was at that very time that he received backing from a16z and @pmarca to develop Uncensored LLMs. a16z supported 8 open-source developers, and Eric Hartford was one of them, alongside @NousResearch and other top devs. - Onchain Research Total Supply: 500M POD Real Circulating: ~39.6M POD (7.9%) Team controls: 444M POD (88.8%) Market Cap: ~$1.46M | FDV: ~$18.4M Team controls most of the supply on Team's Multisig wallet and Treasury wallet. The real circulated supply is too low and easy to pump. If you look at the @nansen_ai list of holders for this token, you’ll see >20 holders who each hold >10k worth of $POD. A dozen whales with >$100K balances. There are 5 wallets with >$1M holding $VVV alongside high-conviction plays like $AERO and $DAI. These aren't low-cap gamblers, they only move on assets with clear rationale. I’m even see one smart holder who got 20x on an under-radar $DONUT ponzi play in December. - Liquidity Pool (2.5M -> 800K) At its peak, the pool had $2.5M in liquidity - deep enough for smart whales and insiders to load up without moving the needle too much. On April 27, volume hit $1M with zero CT mentions - the price only rose 79%, allowing major players to enter unnoticed. Then the shift: on May 2, the team pulled liquidity down to $800K. Now, you have a concentrated group of high-balance conviction holders sitting on a token that's primed for a vertical move. With a lightened pool and a solid thesis. - My view This token is probably the best bet after $VVV on Decentralized AI with real use case. With @AskVenice backing it can go to hunderd of millions in FDV. Models developed by Dolphin has over 4m monthly downloads on Hugging Face. Add to that the following planned updates: - Launch of the Inference API; - Launch of subscriptions with different tiers of access to the services provided on the network. 100% of revenue will buy and burn $POD. When earnings < node rewards, emissions are there to cover the difference. When earnings > node rewards, the protocol is profitable & excess rewards will be distributed to $POD stakers to grow their relative share of the network. Real Product Usage + Backing by Venice + Founder which got grant by a16z + big buys with zero mentions on CT + Revenue model which focused on token growth + Zero Hype yet. More info, proofs and screens you can see in the comments to the tweet. This is the second very detailed research after $ZOE, gonna write up more and more when interesting projects occur.

@AntSeedAI and this from @dphnAI

@0x_bzzz @0xJeff We verify node operators are running the expected model by comparing logprobs + tokenised output to validators & a bunch of other techniques

@0xJeff Its not even verified inference what is the point. Get this nonsense off my feed bro. Trust some random stranger on the internet

Dolphin AI is on a generational run right now ​ People are finally catching up to the inference thesis ​ (Congrats to all the Substack subscribers who got in)

@diegoxyz @0xJeff sent you a dm with an overview of our project

@0xJeff what's their value prop?

Venice Uncensored 1.2 is now live. Developed with @dphnAI, this model delivers the most uncensored version of Mistral 24B. Upgraded with vision support, a 4x larger context window, and stronger tool-use capabilities. Trained on Bittensor Subnet 4 @TargonCompute.

项目方发布了白皮书，还贴心的用gpt翻译了中文版。毕竟排行榜前几位都是认识的国人老哥👍 下周会上线质押合约，可以通过绑定获得奖励倍数。 drive.google.com/file/d/1DnTcpj…

@noboom96814 @cryptoresetlife No

@cryptoresetlife @dphnAI 对网络有严格要求吗？

谢谢官方@dphnAI ！！！ 5天给了7220个 $pod 目前价值250u 之前币价拉到0.045。希望继续 🙏

@coolfun6 Dm wallet address Our anti-cheat automatically banned & slashed a few providers that tried to modify the worker

@dphnAI what's wrong?

First epoch of rewards for node providers has been paid out 50K $POD distributed to 33 providers based on relative contributions to the Qwen 35B inference pool datagen.dphn.ai

@DecentrlizOrDie Nice thread - much more to come 🐬

$POD ( @dphnAI) Timeline They’ve been quietly building for 2.5 years — first as one of the most popular open-source AI model makers, now as a decentralized inference network

@coolfun6 Thrusday this week - epoch is every 7 days

@dphnAI when we could withdraw $POD

Node provider rollout has been going well Our pool of inference nodes running Qwen 3.6 35B have generated over 3.2B tokens so far Total inference bandwidth -> 9400 t/s 28x RTX 4090 12x RTX 5090 8x RTX PRO 6000 & many other cards API access coming soon 🐬

Guide on how to run a node in our docs dphn.ai/docs/running-a… 60gb vRAM required to run in FP8 with full context We recommend 1x RTX 6000 PRO or H100 / H200 / B200 on @TargonCompute Smaller models for idle consumer GPUs coming soon ~~ Watch the 35B datagen live datagen.dphn.ai

@cryptoresetlife @zise772352 We increased utilisation % today Rewards should be a bit higher now

@zise772352 @dphnAI 一天一个1000多

第五名了✌️ 下周打款说好了哈🤝@dphnAI