Arcee.ai

Arcee's Trinity-Large-Thinking is live on its own API, OpenRouter and Hugging Face under Apache 2.0. Two months of SFT and RL work followed Preview, aimed at multi-turn tool use, context coherence across turns, instruction following under constraint, and stable long-running behavior.

Charles Goddard details how AFM-4.5B's context window went from 4k to 64k through heavy experimentation with model merging, distillation and repeated model soups, rather than a single dedicated long-context training run.

Arcee researchers Charles Goddard and Fernando Fernandes Neto describe tokenizer transplantation: using orthogonal matching pursuit to express one model's embeddings in another's vocabulary, so models with different tokenizers can be merged, distilled or paired for speculative decoding without retra

Arcee released Trinity Large as three variants from one training run: a chat-ready Preview, a full 17T-token Base, and a 10T-token TrueBase with no instruct data or learning rate anneals. The MoE routes 4 of 256 experts, trained on 2,048 B300 GPUs in 33 days using z-loss and momentum load balancing.

Arcee is handing out inference credits on its API to open-source projects, research groups, prototypes and tooling builders, sized from under 50M to over 1B tokens per grant. Applications are reviewed on a best-effort basis against available capacity.

Arcee open-sourced DistillKit, the toolkit behind Virtuoso and SuperNova Medius. It runs online and offline distillation from a single YAML config, compresses stored teacher logits to roughly 300 bytes per token, and ships captured teacher data from Qwen3-235B and DeepSeek V3/R1.

Arcee introduced its Foundation Model family, starting with AFM-4.5B, a 4.5 billion parameter model targeted at running on phones, edge hardware or cloud. It is testable in Arcee's playground and on Together AI, with training pipeline and benchmark detail in a companion technical post.

Arcee released open weights for five models: the production systems SuperNova, Virtuoso-Large and Caller that had been serving customer workloads on its SaaS, plus research models GLM-4-32B-Base-32K and Homunculus, as the company shifts focus to its AFM foundation line.

Trinity-Large-Thinking is free on OpenRouter until May 23, positioned for multi-turn tool use and long-horizon routing. Arcee notes the model is tuned to carry reasoning across calls, which OpenRouter documents as interleaved thinking.

Arcee took AFM-4.5B out of preview and published weights for both the instruction-tuned model and AFM-4.5B-Base on Hugging Face, moving its first foundation model from a hosted preview to something teams can download, run across cloud or edge, and fine-tune themselves.

Arcee is building Genesis-Science-1 with the US Department of Energy: an open-weight model paired with a governed research harness meant to run scientific computing workflows while keeping a reproducible record of its work. A contributor program opens to labs and universities.

Arcee is switching off its hosted Trinity-Large-Preview endpoint on April 22, 2026, redirecting constrained serving capacity to paid Trinity-Large-Thinking users. The team says it would work with any outside provider willing to host Preview instead.

Intel's OpenVINO 2025.3 release adds full support for Arcee's AFM-4.5B, and quantized 16-bit, 8-bit and 4-bit builds are published on Hugging Face. That covers Intel CPUs and GPUs from edge devices through cloud instances.

Arcee has moved its Trinity open-weight models onto the OpenMDW-1.1 license, changing the terms attached to the released weights.

nac 0.1.3 groups sessions under projects, isolates each one in its own sandboxed git worktree, adds skill prompt expansion with composer autocomplete, reads images, and renders code and diffs with Shiki highlighting.

Arcee is shutting the hosted Trinity-Large-Preview endpoint on OpenRouter after roughly 3.3T tokens served since January, pointing active pipelines at Trinity-Large-Thinking. The Preview weights stay on Hugging Face for local deployment.

Arcee published pre-anneal base checkpoints for its Trinity Nano and Mini models. They retain the math and code exposure of the released bases but sit earlier in the training schedule, which the team says makes continued pretraining and customization easier than starting from post-anneal weights.

Arcee is pulling the hosted Trinity Mini checkpoint from OpenRouter and reallocating that serving capacity toward the next Trinity generation. Existing workflows must switch models; the Mini weights stay on Hugging Face for local deployment.

Arcee reversed the planned Preview sunset: Trinity-Large-Preview stays live at $0.15 in and $0.45 out per million tokens, while Trinity-Large-Thinking falls from $0.90 to $0.80 per million. Both are reachable through OpenRouter or Arcee directly.

Arcee and Nous Research paired the Hermes agent with Trinity-Large-Thinking, published alongside a walkthrough for getting the harness installed and configured on Linux, macOS and Windows.

Trinity opens Arcee's open-weight mixture-of-experts line with Trinity-Mini at 26B parameters and 3B active over 128 experts, alongside a 6B Nano preview with 1B active. Weights are on Hugging Face, with a free access window on OpenRouter at a 131k context window.

Arcee engineers benchmark AFM-4.5B on Intel Xeon, AWS Graviton and Qualcomm CPUs to establish which configurations make CPU-only serving viable, covering the hardware and software improvements that close the gap for cloud and edge deployments.

After close to 3 trillion tokens in 50 days, Arcee is applying what it calls lenient rate limits to the free Trinity Large Preview endpoint. Free access remains for now, with added capacity and a paid endpoint planned for workloads that need consistent throughput.

OpenClaw's 2026.4.7 release adds a bundled provider plugin spanning Arcee's full model catalog and OpenRouter, making Trinity-Large-Thinking a dropdown selection rather than a custom endpoint configuration.

Loka post-trained Arcee's 26B Trinity Mini into a science research harness: slash commands trigger fixed workflows, an internal critic stress-tests hypotheses against retrieved evidence, and ambiguous queries resolve identifiers against PubMed, UniProt and GEO before concluding.

Arcee's hosted API now serves GLM-5.3 and GLM-5.3-Flash, extending a catalog it opened two weeks earlier beyond its own Trinity line to DeepSeek, Kimi and Inkling. All of them sit behind a single endpoint and are callable from its nac agent for long-running, hands-off work.

Mergekit is now licensed under GNU LGPL v3, a change Arcee frames as returning the model-merging toolkit to its original terms and clearing up what users and contributors are permitted to do with it.

Arcee reports Trinity-Large-Thinking at 91.9 percent on PinchBench, the OpenClaw agent benchmark from Kilo Code, second to Claude Opus 4.6 at 93.3 percent. Input runs $0.25 per million tokens with cache reads at $0.06, and OpenRouter made it free for OpenClaw through April 5.

nac is an Apache 2.0 harness where an orchestrator dispatches work to isolated workers; each worker exits and drops its raw context, leaving a summary episode other threads can weave in. It ships an MCP server, alongside an Arcee API beta hosting several frontier open models.

NAC v0.1.4 adds conversation forks: a finished chat prefix clones into an independent session keeping workspace, model, threads and steering, without duplicating the sandbox worktree. Stop now confirms model, tool, worker, MCP and command teardown before it reports a run cancelled.