narrow _exclude to technical incapacity only

narrow _exclude to technical incapacity only

AppImage/config/verified_ai_models.json

@@ -1,7 +1,7 @@
 {
   "_description": "Verified AI models for ProxMenux notifications. Only models listed here will be shown to users. Models are tested to work with the chat/completions API format.",
-  "_updated": "2026-07-14",
-  "_verifier": "Refreshed by .github/workflows/verify-ai-models.yml (daily). The workflow runs .github/scripts/ai-models-verifier/verify.py against every provider whose API key is configured in repository Secrets, then applies the report via apply.py — which honours per-provider `_exclude` lists so editorial curation survives automated refreshes. Manually re-run from the Actions tab (any branch) when a new model needs to be picked up out of cycle.",
+  "_updated": "2026-09-02",
+  "_verifier": "Refreshed by .github/workflows/verify-ai-models.yml (daily). The workflow runs .github/scripts/ai-models-verifier/verify.py against every provider whose API key is configured in repository Secrets, then applies the report via apply.py — which honours per-provider `_exclude` lists. `_exclude` is intentionally minimal: it only drops models that are technically incapable of generating a chat completion for a normal prompt (safety classifiers, agentic-only endpoints, meta-routers, wrong modalities). Everything else the verifier passes is surfaced — including language-specialised models (Arabic, Chinese, ...), reasoning models, legacy families and dated snapshots — so a user with a specific need can still pick the model that fits. Manually re-run from the Actions tab (any branch) when a new model needs to be picked up out of cycle.",
 
   "groq": {
     "models": [
@@ -13,14 +13,11 @@
     ],
     "recommended": "llama-3.3-70b-versatile",
     "_exclude": [
-      "allam-2-7b",
-      "qwen/qwen3-32b",
-      "qwen/qwen3*",
       "openai/gpt-oss-safeguard-*",
       "groq/compound",
       "groq/compound-*"
     ],
-    "_note": "Verified functionally 2026-07-14 with the Groq API. `_exclude` covers models that pass the technical test but are the wrong fit for notification translation: allam-2-7b (Arabic-focused), qwen/* (Chinese-first, unreliable Spanish), openai/gpt-oss-safeguard-* (safety-classifier variant), groq/compound* (agentic system, not a chat model)."
+    "_note": "`_exclude` covers models the verifier may technically pass but that do not produce a usable chat completion: openai/gpt-oss-safeguard-* is a safety classifier (returns a category, not free text); groq/compound* is an agentic system that expects multi-step tool use, not a plain prompt."
   },
 
   "gemini": {
@@ -34,16 +31,12 @@
     ],
     "recommended": "gemini-2.5-flash-lite",
     "_exclude": [
-      "gemini-*-pro*",
-      "gemini-*-thinking*",
       "gemini-embedding-*",
-      "gemini-2.0-*",
-      "gemini-1.5-*",
-      "gemini-1.0-*",
-      "gemini-pro"
+      "gemini-*-pro*",
+      "gemini-*-thinking*"
     ],
     "_deprecated": ["gemini-2.0-flash", "gemini-2.0-flash-lite", "gemini-1.5-flash", "gemini-1.0-pro", "gemini-pro"],
-    "_note": "Verified 2026-07-13. gemini-flash-lite-latest now passes consistently (1.6s) and is fastest, but gemini-2.5-flash-lite remains recommended because 'latest' aliases can drift over time. gemini-3.1-flash-lite is the stable successor to 3-flash-preview. Pro variants continue to reject thinkingBudget=0 and are overkill for notification translation."
+    "_note": "`_exclude` drops embeddings (wrong modality) and Pro / thinking variants that reject `thinkingConfig.thinkingBudget: 0` and therefore never return a visible completion within a reasonable token budget — technical failure with our current provider config."
   },
 
   "openai": {
@@ -58,11 +51,6 @@
     ],
     "recommended": "gpt-4.1-nano",
     "_exclude": [
-      "gpt-3.5-*",
-      "gpt-3-*",
-      "gpt-4",
-      "gpt-4-0613",
-      "gpt-4-turbo*",
       "gpt-4o-audio*",
       "gpt-4o-realtime*",
       "gpt-4o-search*",
@@ -72,22 +60,9 @@
       "gpt-4o-mini-search*",
       "gpt-4o-mini-transcribe*",
       "gpt-4o-mini-tts",
-      "gpt-4.1-nano-2*",
-      "gpt-4.1-mini-2*",
-      "gpt-4.1-2*",
-      "gpt-4o-2*",
-      "gpt-4o-mini-2*",
-      "gpt-5-nano-2*",
-      "gpt-5-mini-2*",
-      "gpt-5-chat-2*",
-      "gpt-5-2*",
-      "o1-*-2*",
-      "o3-*-2*",
-      "o4-*-2*",
-      "codex-*",
       "computer-use-*"
     ],
-    "_note": "Verified 2026-07-13. `_exclude` drops (a) dated snapshots (`gpt-4o-2024-11-20`, `gpt-4.1-nano-2025-04-14`, ...) — the stable aliases are preferred so the recommended model doesn't silently pin to a specific point-in-time build; (b) legacy families (gpt-3.5, gpt-4, gpt-4-turbo) that gpt-4.1 supersedes; (c) audio/realtime/search/transcribe/tts variants (wrong modality for notifications); (d) reasoning models (o-series, gpt-5.1/5.2 non-chat), which openai_provider.py supports via max_completion_tokens + reasoning_effort=minimal but do not improve translation quality and are slower. Add specific reasoning IDs to `models` manually if a user explicitly wants them."
+    "_note": "`_exclude` covers wrong-modality variants (audio, realtime, search, transcribe, tts) that cannot handle a plain notification-translation prompt, plus computer-use which requires an agent loop. All other OpenAI chat/completion models are surfaced — including legacy families (gpt-3.5, gpt-4), reasoning models (o-series, gpt-5.x non-chat) and dated snapshots — so users can pick by their own criteria (cost, quality, reproducibility). openai_provider.py already handles reasoning models via max_completion_tokens + reasoning_effort=minimal."
   },
 
   "anthropic": {
@@ -100,13 +75,8 @@
       "claude-fable-5"
     ],
     "recommended": "claude-haiku-4-5",
-    "_exclude": [
-      "claude-*-2*",
-      "claude-3-*",
-      "claude-3-5-*",
-      "claude-3-opus-*"
-    ],
-    "_note": "Verified 2026-07-13 with all 10 discovered models passing after aligning the verifier with anthropic_provider.py (temperature omitted — newest generations reject it with 'temperature is deprecated for this model'). `_exclude` drops dated snapshots (`claude-haiku-4-5-20251001`) and legacy generations (claude-3-*) that are deprecated upstream. haiku-4-5 is the sweet spot for notification translation (3.6s, $1/$5 per MTok); sonnet-5 for slightly richer output (3.1s, $3/$15); opus-4-8 / fable-5 for demanding cases."
+    "_exclude": [],
+    "_note": "No technical exclusions — every Claude generation returns free-text completions for a plain prompt. The verifier tests each model listed under `models`; if a specific ID stops working upstream it simply drops out of the passing set. Anthropic does not expose a public models-list API, so new models must be added to `models` manually before the verifier can test them."
   },
 
   "openrouter": {
@@ -131,32 +101,17 @@
     ],
     "recommended": "meta-llama/llama-3.3-70b-instruct",
     "_exclude": [
-      "*/wizardlm-*",
-      "*/qwen*",
-      "*/yi-*",
-      "*/ernie-*",
-      "*/glm-*",
-      "*/deepseek*",
-      "*/hermes-*-405b*",
-      "*/dolphin*",
-      "*/euryale*",
-      "*/mythomax*",
-      "*/toppy*",
-      "*/rocinante*",
-      "*/nsfw*",
-      "*/*-uncensored*",
-      "*/*-vision*",
-      "*/*-image*",
+      "openrouter/*",
       "*/*-audio*",
+      "*/*-audio-*",
       "*/*-tts*",
       "*/*-whisper*",
       "*/*-embed*",
-      "openai/o1-*",
-      "openai/o3-*",
-      "openai/o4-*",
-      "anthropic/claude-3-*"
+      "*/*-embedding*",
+      "*/*-image*",
+      "*/*-vision-only*"
     ],
-    "_note": "OpenRouter aggregates hundreds of models — `_exclude` is aggressive by design so the surfaced list stays curated. Blocked families: Chinese-first (Qwen, Yi, ERNIE, GLM, DeepSeek), role-play / uncensored, wrong-modality (vision/image/audio/tts/whisper/embed), reasoning models (o-series) and legacy Claude 3. Kept: the mainline chat/instruct models that were manually validated. Paid tier verified functionally 2026-07-14; free tier verified 2026-08-17. Free-tier rate limits (~20 req/min shared across all OpenRouter free users on that model) may cause 429 in high-traffic windows — usable for occasional notification translation, not for high-volume automation."
+    "_note": "OpenRouter aggregates hundreds of models; the free-tier variants (:free suffix) are intentionally supported per user request and never blocked. `_exclude` covers only meta-routers (`openrouter/free`, `openrouter/auto` — they route dynamically to something else, so their behaviour is not the model the user picked) and wrong-modality models (audio, tts, whisper, embeddings, image, vision-only). Everything else — chat models across every family, language and price tier — is surfaced so the user can pick the fit."
   },
 
   "ollama": {