All Posts
All Posts
BottleCap AI
Sep 25, 2026 · 3 min read
Halve the thinking trace length, quadruple the work
Shorter reasoning is cheaper and faster, but on a server that is limited by KV cache it gets even better. Cut the length of each answer by 2.3× and the same GPU can finish 4.7× as many tasks within the same time window.
BottleCap AI
Sep 22, 2026 · 12 min read
ThinkingCap-Qwen3.8-27B: the same answers, 37% less thinking
The second model in our ThinkingCap series. We took Qwen3.8-27B and cut how much it thinks, without changing the quality of the answers.
BottleCap AI
Jul 13, 2026 · 7 min read
Introducing efficiency focused “ThinkingCap” model series.
One of our core aims at BottleCap is efficiency in AI. To fulfil this mission, we fine-tuned Qwen3.6-27B to reduce unnecessary reasoning while preserving answer quality.
BottleCap AI
May 2, 2026 · 1 min read
BottleCap AI Named Among 21 European Startups to Watch by TechCrunch
TechCrunch has included BottleCap AI in its curated list of 21 European startups worth knowing. The Prague-based startup was personally recommended by Julien Codorniou, general partner at 20VC — as a direct investment from his portfolio.
BottleCap AI
May 13, 2025 · 1 min read
One Month In: We've Already Cut LLM Training Costs by 50%
One month ago, we started BottleCap AI with the mission to build a European research hub to make LLMs (today's AI models) radically more efficient with the ambitious goal to increase efficiency of most prominent models by 100x.
BottleCap AI
Jan 29, 2026 · 2 min read
We raised $7.5M to build efficient AI in Europe
We're excited to announce that we've raised a $7.5M seed round led by 20VC alongside an incredible group of founders (Canva, ElevenLabs, Datadog, Huggingface, Lovable, Synthesia, Pigment, and many more), to bring efficient, architecture-first foundational LLMs to the world.