/models REST endpoint lacks context size

[rofl0r]

Describe the bug
the models endpoint only shows the configured model name and a timestamp. the lack of a context window size makes it impossible for agentic software to know when to trigger auto-compaction.

To Reproduce

$ curl ovms_server:8081/v1/models
{"data":[{"id":"Ornith-1.5-35B-A3B-Uncensored-OpenVINO-INT4","object":"model","created":1790460054,"owned_by":"OVMS"}],"object":"list"}

Expected behavior

$curl llamacpp_server:8081/v1/models
{"object":"list","data":[{"id":"/home/user/models/offmonreal-Ornith-1.5-35B_Q4_K_M_imatrix_MTP.gguf","object":"model","created":1790459973,"owned_by":"llamacpp","meta":{"vocab_type":2,"n_vocab":248320,"n_ctx_train":262144,"n_embd":2048,"n_params":34660610688,"size":21155768832},"max_model_len":262144}],"models":[{"slug":"/home/user/models/offmonreal-Ornith-1.5-35B_Q4_K_M_imatrix_MTP.gguf","display_name":"/home/user/models/offmonreal-Ornith-1.5-35B_Q4_K_M_imatrix_MTP.gguf","description":null,"default_reasoning_level":null,"supported_reasoning_levels":[],"shell_type":"default","visibility":"list","supported_in_api":true,"priority":0,"additional_speed_tiers":[],"service_tiers":[],"default_service_tier":null,"availability_nux":null,"upgrade":null,"base_instructions":"","model_messages":null,"supports_reasoning_summaries":false,"default_reasoning_summary":"auto","support_verbosity":false,"default_verbosity":null,"apply_patch_tool_type":null,"web_search_tool_type":"text","truncation_policy":{"mode":"tokens","limit":262144},"supports_parallel_tool_calls":false,"supports_image_detail_original":false,"context_window":262144,"max_context_window":262144,"auto_compact_token_limit":235929,"effective_context_window_percent":95,"experimental_supported_tools":[],"input_modalities":["text"],"supports_search_tool":false,"use_responses_lite":false,"auto_review_model_override":null,"tool_mode":null,"multi_agent_version":null}]}

Logs
Logs from OVMS, ideally with --log_level DEBUG. Logs from client.

Configuration
2026.04 from last week
https://huggingface.co/vonvonhero/Ornith-1.5-35B-A3B-Uncensored-OpenVINO-INT4

Additional context
Add any other context about the problem here.

[rofl0r]

for the record: the other half needed for proper context window accounting is usage statistics like llama.cpp sends them on the last SSE fragment of each turn, or alternatively the completion_tokens value in each SSE packet, so one can accurately calculate the current context window usage. i'll gladly open another bug request if so desired, but if it can be fixed without additional bureaucracy, even better.