@bitrefill First test done.
Now I am going to directly integrate with Hermes, which is helping me furnish my new condo. Several gift cards she'll be able to leverage to buy the stuff I need. ๐ค
@Makintern The Dashboard is a gem.
I cannot stress enough how important it is to have the certainty that AUM / NAV is properly and regularly updated.
E.g: one can easily see that Sr Royco was updated 31 min ago.
@0xDovah@Dialectic_Group@AskVenice My only comment is that some models are heavily politically-correct / biased on certain topics, I dont know if any of this would affect the type of bets in Polymarket.
Then again, that is what @AskVenice is for ๐ซก
@0xDovah "A swarm of onchain agentic judges using reliable sources to find a neutral resolution would solve this".
Agreed with you... maybe like @Dialectic_Group does with the Defi-Bench, that is tapping into 8 different LLM models via @AskVenice?
Agentic payments only make sense on crypto rails. And Bitrefill is THE undisputed crypto shopping mall.
Put them together and that's what agentic commerce actually looks like.
Come test it yourself in Berlin during @BerBlockWeek.
@Dialectic_Group As my credit card could tell you, Claude will clearly not win at being the least expensive.
Let's see if at least it can now bag a metal...๐ฅ๐ฅ๐ฅ or I'll need to consider switching.
@Dialectic_Group Honestly some senior tranches are fantastic, e.g syrupUSDC losing only 20bps with the normal underlying asset, or sNUSD losing 60bps.
I really like the Risk:Reward ratio of the vault and their underlying components.
Qwen 3.7-max beats Opus 4.7 and GPT-5.5
We tested three frontier models on a real agentic task: write a Tetris bot that plays the game and trains itself. Each model could read its own code, run benchmarks, and rewrite itself across 10 iterations. Then we compared the final bots head to head.
Qwen 3.7-Max: training cost $1.32, bot improvement +56%
Claude Opus 4.7: training cost $12.15, bot improvement +28%
GPT-5.5: training cost $2.85, bot improvement +7%
Qwen won on every dimension - biggest jump, 9ร cheaper than Claude, 2ร cheaper than GPT. Long agentic loops is where Qwen Max actually delivers.
@Stronghandsinat Hats off with this concise but brilliant document!
I have already committed $5k to the winning model in the end by the way ๐คฃ
https://t.co/lihuKsr4Sw
@drowrangerxyz@Stronghandsinat oof it is a rude awakening for western people but Swen by Alibaba is waaaay cheaper and seems to be better at several tasks.
https://t.co/fjCf2q6lar
Qwen 3.7-max beats Opus 4.7 and GPT-5.5
We tested three frontier models on a real agentic task: write a Tetris bot that plays the game and trains itself. Each model could read its own code, run benchmarks, and rewrite itself across 10 iterations. Then we compared the final bots head to head.
Qwen 3.7-Max: training cost $1.32, bot improvement +56%
Claude Opus 4.7: training cost $12.15, bot improvement +28%
GPT-5.5: training cost $2.85, bot improvement +7%
Qwen won on every dimension - biggest jump, 9ร cheaper than Claude, 2ร cheaper than GPT. Long agentic loops is where Qwen Max actually delivers.