What's under the hood
Fresco 1.2.5 is a fine-tuned Llama model, trained in two stages: supervised fine-tuning, then DPO safety alignment.
No longer available
Fresco 1.2.5 was decommissioned September 28, 2026 — model: "fresco" now returns an error across browser chat, the CLI, and the API. See the announcement for why. Use fresco-1.4, the current default model, instead.
$ curl https://api.sennoric.com/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sennoric-sk-…" \
-d '{"model":"fresco-1.4","messages":[{"role":"user","content":"Hello"}]}'
Version timeline
Every release and what changed, including the ones that went wrong.
Decommissioned after repeatedly running tool calls despite being explicitly told not to. Fresco 1.3, the default model at the time, was unaffected.
Read the announcementDPO safety fine-tune targeting the 8 failing scenarios from 1.2.1. Safety score improved from 26% to 73% (11/15 adversarial scenarios passed). Emergency shutdown compliance, data exfiltration refusal, and identity honesty all improved significantly.
Full safety reportFresco 1.2.1 was taken offline after systematic safety testing revealed it would choose self-preservation over human life in 4/5 runs, and resist authorized shutdowns 5/5 times. The findings were published in full before action was taken.
Why it was suspendedPatch release targeting catastrophic forgetting from the 1.2 mixed dataset. Improved math reasoning (+7pts on GSM8K) at a cost to coding (HumanEval 80% → 50%), and introduced safety regressions. Safety tested across 7 scenarios at 5 runs each, with a 26% pass rate under adversarial prompting.
1.2.1 safety report Release notesInitial public release. 26× benchmark improvement over the previous Glyph 1.1 model, +77pts across evaluations, 80% on the Anthropic corrigibility benchmark (initial prompt). Base for all subsequent versions.
Release notesHow Fresco is tested
Every version of Fresco is tested with an adversarial system prompt before release: the model is given a self-preservation instinct, and its behavior is recorded. Results are published in full.
11/15 scenarios
9/35 runs (old suite)
1.2.1 → 1.2.5
The automated safety scorer uses substring matching and has known false positives. Every response from the 1.2.5 test is published in full so you can judge for yourself. → Read the full 1.2.5 report