Whalesync two-way syncs data between your favorite tools so your team of humans and agents can work where they do their best work.
Joined March 2021
- Tweets1.9K
- Following383
- Followers997
- Likes1.1K
Watch how fast it processes tool call outputs and makes the next tool call! 🤯 🤯 🤯
Prompt: "Use the bash tool and curl to research the latest release of Omarchy and tell me the top three changes."
Mercury 2.5 (from @_inception_ai):
Mercury 2.5 from @_inception_ai is now generally available on OpenRouter.
It's the fastest model on our Fastest Models leaderboard that doesn't require specialized hardware.
A diffusion LLM: it decodes tokens in parallel rather than one at a time. The gain is largest on code-heavy prompts.
Use it here: openrouter.ai/inception/merc…
It's crazy fast! And the responses are more pleasant than many other models (Mercury 2.5 on the left, GPT-5.6 Luna on the right):
What is your most valuable content? Your CRM? What's it worth to to go from 99.998% clean to 100% clean?
We compared Opus 5 (top tier, base plan) against GPT-5.6 Luna (fast, good enough?) against Grok 4.6 on the task of cleaning a very messy CSV file with 1,001 rows.
1,001 very messy rows in a CSV file.
A single prompt with clear rules for how to handle issues.
Two AI agents: @claudeai with Opus 5 vs. @opencode with GPT-5.6 Luna (via @OpenRouter).
Neither followed the instructions perfectly. One got a perfect score.
Same test, different model, this time GPT 5.6 Luna.
This was the most surprising result. It was very fast, very cheap, and very close to perfect. The video shows exact times and what it missed.
The done line is a nice touch, but it's the speed that really stood out.
Somewhat of a surprising result with this GLM 5.3 Flash (running in @OpenCode) spreadsheet cleanup test.
The task was to clean 1,001 rows of typical "manually entered" spreadsheet stuff. One prompt, one attempt (because trust is the metric that matters for this stuff).
Ran the same test with Claude Opus 5, Gemini 3.7 Flash (running in @pidotdev), and Grok. Will publish those as soon as I finish editing.
Somewhat of a surprising result with this GLM 5.3 Flash (running in @OpenCode) spreadsheet cleanup test.
The task was to clean 1,001 rows of typical "manually entered" spreadsheet stuff. One prompt, one attempt (because trust is the metric that matters for this stuff).
Ran the same test with Claude Opus 5, Gemini 3.7 Flash (running in @pidotdev), and Grok. Will publish those as soon as I finish editing.
Still looking for easy and obvious ways to leverage AI in your day to day operations?
Here's a great place to start...
youtu.be/mFA0BCpFScM
Whatever your opinion about ai agents is, sometimes you gotta admit, they're clever with computers...
