New: AI API Cost Calculator — compare GPT, Claude, Gemini, Grok, Llama, DeepSeek & Kimi pricing live
Developers · News

K2-Horizon Opened Everything: 375B Open Weights, Data and All

By ToolsDig Research · Updated November 17, 2026 · How we research

Nothing here is sponsored — no affiliate links on this page. Every fact below comes from the K2-Horizon paper (arXiv:2611.21003, submitted November 16, 2026) and IFM's page at ifm.ai/k2. We have not run these models hands-on — this is researched analysis, not hands-on testing. Benchmark claims are the paper's own and have not been independently verified.

A sleek modern workspace at dusk, a laptop screen glowing with neural network diagrams and lines of code, a large open glass box on the desk containing glowing translucent layers

Most "open" model releases give you the weights and wish you luck. IFM just did something I haven't seen at this scale: they opened the weights, the training data, the intermediate checkpoints, the training logs, the infrastructure code, and the pipelines that built the data. All of it. That's the story here — not another model, but a release that treats reproducibility as the product. On November 16, 2026, IFM (Abu Dhabi) published the paper for K2-Horizon (arXiv:2611.21003), a fleet of six fully open language models, from a 0.9B dense checkpoint up to a 375B-A23B mixture-of-experts flagship built for coding and agentic work.

🔍 How we put this together

We used the K2-Horizon paper (arXiv:2611.21003, submitted November 16, 2026) and IFM's page at ifm.ai/k2. Where a figure is the paper's own claim — benchmark competitiveness, the 2.2x decoding speedup, the token counts — we say so. Nothing here is independently verified yet, and we have not run the models ourselves.
🏆 Verdict: The weights are the least interesting part of this release. What matters is the full development history: checkpoints from every training stage, the pretraining data, the training logs, the xLLM and RL360 code, and the documented data pipelines behind roughly 10T synthetic tokens. If you build with open models, start with the 7.0B or the 36B-A4B, read the data pipeline docs before fine-tuning, and test Uno on a real agentic workload.

The fleet, in plain terms

Six models, one shared design language, sized for different jobs:

Four of the six share one 22-trillion-token pretraining corpus and recipe — the kind of controlled setup researchers dream about: same data, same recipe, different scales and sparsity, so you can actually study what scale does instead of guessing across releases that changed everything at once.

The models are built for general-purpose reasoning and agentic work — math and science, software engineering, deep research, office automation. The paper says they lead or are competitive with strong open models of their size on agentic and coding benchmarks. That's the paper's claim, not ours, and we won't dress it up with numbers we haven't verified.

Three technical ideas worth knowing

MoVA — Mixture-of-Value Attention. The paper applies mixture-of-experts thinking to the attention value projections. The 36B-A4B model uses it, and the headline result is that it nearly matches a 32B dense model on the same data with roughly one-seventh the active parameters. If that holds up in independent testing, the small-MoE shape gets a lot more interesting for anyone serving models on a budget.

Softplus attention gating. A more flexible, theoretically motivated gate on attention, also in the 36B-A4B. What matters practically is that it's part of why the small MoE punches above its active parameter count.

Uno — diffusion adapters for faster decoding. Uno is a lossless diffusion adapter that speeds up decoding by up to 2.2x, according to the paper, without needing a separate draft model. Speculative decoding usually means maintaining a second small model; Uno skips that. If you're serving an agent that makes dozens of sequential tool calls, decoding speed is the tax you pay on every step — a 2x speedup there is worth more than a few benchmark points.

The real news: they opened the kitchen, not just the dish

Here's what makes this release different from every other "open weights" drop this year. IFM says the release follows the LLM360 definition of open source for transparency and reproducibility, and the list is long:

That last one is the sleeper. Those pipelines produced roughly 10 trillion synthetic pretraining tokens. Synthetic data at that scale, with the pipeline open, means other labs can actually study how synthetic data behaves as a major pretraining ingredient instead of theorizing about it. IFM calls it a glass box, not a black box, and for once the phrase fits the release.

This is the difference between "you can download our model" and "you can see how we built it, reproduce it, and build on it." Most open releases stop at the first sentence. K2-Horizon tries for the second.

The honest limits

What to do this week

If you build with open models, the move is to start small and inspect, not to chase the flagship:

1. Pull the 7.0B or the 36B-A4B first. The 7B is the cheap experiment; the 36B-A4B is the interesting one if you want to see whether MoVA delivers in practice.
2. Read the data pipeline docs before you fine-tune. The pipelines are the part nobody else publishes. Even if you never retrain, knowing how the synthetic data was built tells you what the model is likely good and bad at.
3. Try Uno on an agentic workload. If you have an agent loop that makes many sequential calls, a decoding speedup without a draft model is a real serving win — test it on your own latency, not the paper's number.
4. Compare the 36B-A4B against the 32B dense on your own tasks. Same data, same recipe, one-seventh the active parameters — that's a clean experiment, and it's the claim most worth checking yourself.

Sources: K2 Team, "K2-Horizon: A Frontier Fleet of Open Source Language Models," arXiv:2611.21003, submitted November 16, 2026; IFM, "K2 Horizon: Open-Source AI Models for Every Scale," ifm.ai/k2. Model sizes, architecture details, token counts and openness claims are from the paper and IFM's page; benchmark claims are the paper's own and have not been independently verified.