what we learned to keep coding agents fast in Claude Code, Codex, Pi, and more:
– a single timestamp in the system prompt can break the cache
– changing reasoning effort can wipe the whole cache
– telling the agent its output was cut off will get it to read what it missed
coding agents resend the whole conversation on every turn. prefix caching keeps long sessions fast, but only if the harness and the inference engine both do their part.
@zainhas from Together and @parthsareen from @ollama wrote down what each side needs to get right.
together.ai/blog/making-open…
Inkling 2 Small from Thinking Machines (@thinkymachines) is now on Ollama with zero data retention!
The 594B mixture-of-experts model with 1M context window demonstrates strong visual interpretation and creation, agentic knowledge work, and finance and quantitative reasoning.
Example use cases:
- OCR
- grounding
- infographic interpretation
- visual reasoning
- tool calling with MCP
- finance spreadsheet reasoning, filings and documents
Available on Ollama Cloud with zero data retention:
ollama run inkling-2-small:cloud
Roughly 150 developers joined us and @friendliai for a meetup in Seoul last week, where our CEO @jmorgan talked about the open model ecosystem and where it’s headed. Huge thank you to everyone in the community who showed up – more than 650 registered, and we wish there was room for all of you.
We made @theinformation list of top 50 promising early-stage startups for 2026 🎉
Thank you to everyone who runs, builds on, and talks about Ollama.
theinformation.com/articles/…
It was an honor to host South Korea’s top tech leaders for a dinner in Seoul with our partners @friendliai. Over 100 executives joined us to discuss the future of open models and how we can collaborate with enterprises and developers across Korea.
ollama retweeted
오늘 @ollama , @friendliai , @MiniMax_AI 주관 Seoul Meetup 이 있었습니다. 자주 이런 행사가 한국에 더 많이 생겼으면 좋겠습니다.
@jmorgan 의 ollama demo는 ollama cloud 구독자로서 재밌고 신기하기도 하였습니다.
@friendliai 의 기술력이 실제로 실무에 쓰는 vllm에 사용된 기술의 원조라는게 놀라웠고 한국인으로서 자랑스러웠습니다.
@MiniMax_AI 의 다가올 신규 모델 minimax m3 pro 또한 저에게는 깜짝소식었는데 어서 나오길 기대합니다.
ollama retweeted
AI is using more and more of the world’s energy. But how much economic value are we getting from AI?
We’re introducing Joules-to-GDP (J2G), a framework that measures how well AI turns energy into economic value and shows which tasks, models, and hardware give the best return.
Across 44 occupations in finance, engineering, law, administrative work, and more in @OpenAI's GDPval, we arrive at three findings:
1) The tasks with the highest economic value are not the tasks with the highest economic value per joule. For example, CAD design earns ~10% more per task than building websites, but the latter earns ~27% more value per joule.
2) The most powerful model isn’t always the best use of energy. Gemma 4 31B delivers 56% of GLM 5.2’s net value with 23% of the energy, for 2.5x the value per joule.
3) More chips don’t guarantee proportionate returns. Moving the same model from 1 to 8 NVIDIA H100s uses 3.2x the energy but cuts completion time by only 22%, with about the same value.
With @SnorkelAI, we’re also releasing GDPWorks, a benchmark for economically valuable tasks across 36 additional occupations.
Joint work co-led by @amirzeinaalii and @Avanika15, with @Mersad_Abbasi, @Jon_Hartley_, John Hennessy, @HazyResearch, and @Azaliamirh.
More details below 🧵
Kimi K3.1 Pro from @Kimi_Moonshot is now available on Ollama Cloud.
🔒 Hosted in the US with full zero data retention.
2.8T total parameters with 104B activated, designed for coding, 3D creation, financial research, and other agentic tasks.
Run it on Ollama Cloud:
ollama run kimi-k3.1-pro:cloud
Kimi K3.1 Pro rolling out on Ollama Cloud shortly.
US based. Zero data retention.
Releasing the model weights of Kimi K3.1 Pro: huggingface.co/moonshotai/Ki…
We'll keep advancing open intelligence.
We sat down with South Korea’s Vice Chair of AI, @JungWooHa2, in Seoul this week to talk about open source adoption, the critical role open models play in cybersecurity and broadening AI access throughout Korea, and the importance of our partnerships across the Korean AI ecosystem to grow our community.
@jmorgan
ollama retweeted
Some numbers here - @ollama...
Model Warm TPS Cold TTFT
gemma4:e4b 122.8 8.65s
qwen3-coder:30b 90 10.25s
gemma4:26b-a4b 86.9 13.03s
gpt-oss:20b 60.7 9.93s
qwen3.8:latest 25.3 14.84s
It feels really quick for a laptop running local models.
Our CEO @jmorgan was invited to @SeoulNatlUni to give a lecture on open model adoption for more than 200 students. Professors Jae W. Lee, Tae-Wan Kim and the entire team at SNU’s AI Institute are doing incredible work to advance Korea as a global AI leader. Thank you for hosting us!
































