Loading post…
Together AIBenchmark
Kimi K2.7 Code vs Claude Fable 5: Landing pages that cost 94% less
Summary
The experiment compares Kimi K2.7 Code and Claude Fable 5 across 12 landing-page prompts covering B2B SaaS, a rooftop speakeasy, developer tools, and other categories. Kimi cost about 94% less overall, averaging roughly 16 times cheaper than Fable, while producing similar-quality output, especially when given visual references. The team first established a prompt-only baseline, then used a custom design MCP server to provide screenshots, UI elements, and other references directly to multimodal Kimi; this improved hierarchy, typography, composition, readability, loading behavior, and image handling. GPT-5.5 scored screenshots and source code on a 0–100 rubric, and although Fable scored higher, the relatively small quality gap led the authors to consider Kimi’s lower cost a reasonable trade-off for iterative landing-page generation.
Context
The comparison covers 12 generated landing pages and evaluates more than code alone, including positioning, visual direction, content structure, craft, responsiveness, and technical execution. The pages were scored by GPT-5.5, and the experiment tracked costs, token usage, and generation time.
Approach / What changed
The team gave both models the same landing-page prompts, first measuring results from prompts alone. It then supplied Kimi K2.7 Code with screenshots, UI elements, and visual references through a custom design MCP server. GPT-5.5 evaluated each page’s screenshots and source code against a 0–100 rubric.
Takeaways
Kimi K2.7 Code generated the landing pages at about 94% lower overall cost than Claude Fable 5, averaging roughly 16 times cheaper.
A custom design MCP server gave multimodal Kimi screenshots and UI references, producing stronger hierarchy, typography, composition, readability, and fewer broken-image placeholders.
Claude Fable scored higher overall, but the reported quality gap was relatively small compared with Kimi’s cost advantage for iterative page generation.
FAQ
What changed when Kimi received design references through an MCP server?
The model could use screenshots, UI elements, and other visual references alongside the text prompt. The resulting pages had stronger hierarchy, better typography, more intentional composition, faster loading, and fewer broken-image placeholders.
How much cheaper was Kimi for the landing-page experiment?
The average Kimi landing page was roughly 16 times less expensive than one generated with Claude Fable 5. One B2B SaaS page cost 4 cents with Kimi versus $1.09 with Fable, and 100 pages would save around $94.
Did Claude Fable produce higher-quality landing pages?
Claude Fable scored higher in the comparison, but the gap was relatively small. Kimi remained competitive on design, structure, and overall page quality, so the authors considered the lower cost a reasonable trade-off.
Source and freshness
- Source
- together.ai
- Published
- Added to Yomu
Summaries are generated from the original by a language model. Check the original before relying on it.
Keep reading
Related reading
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
The best engineering reads, once a week
One useful email. Unsubscribe anytime.