Refs librefang/librefang#4842 — long-term replacement for the substring
match that the OpenAI driver currently uses to decide how to handle
`reasoning_content` on historical assistant turns.
Three provider-specific behaviours that the driver must distinguish at
wire time, now expressed as catalog metadata:
* `strip` — DeepSeek R1 / deepseek-reasoner. The API rejects requests
that carry reasoning_content on previous assistant messages.
* `echo` — DeepSeek V4 Flash. Thinking mode is on by default and the
API rejects multi-turn requests when assistant turns containing
tool_calls don't echo back the original reasoning text. This is the
bug surfaced in librefang/librefang#4842.
* `empty_string` — Moonshot / Kimi K2 family. The field must be present
(empty string) on tool_calls turns, with thinking disabled wire-side
for multi-turn compatibility.
* `none` (default) — most providers; field is omitted entirely.
V4 Pro is intentionally NOT marked `echo` — librefang#4842 reports it
working out-of-the-box; flip when there's an empirical reproducer.
Marks affected models:
providers/deepseek.toml
deepseek-v4-flash → echo
deepseek-reasoner → strip
providers/moonshot.toml
kimi-k2.6, kimi-k2.5, kimi-k2 → empty_string
providers/kimi-coding.toml
kimi-for-coding → empty_string
providers/byteplus-coding.toml
kimi-k2.5 → empty_string
providers/novita.toml
moonshotai/kimi-k2-thinking → empty_string
Tooling:
* schema.toml registers the field with the four enum options and a
`none` default so existing TOML files keep parsing unchanged.
* scripts/validate.py rejects unknown enum values; verified with a
hand-crafted negative case (`reasoning_echo_policy = "bogus"` →
validation fails with the expected message).
* `python3 scripts/validate.py` passes (267 models).
The librefang side that consumes this field will land in a follow-up
PR — until then, registry consumers ignore the field via
`#[serde(default)]` and the existing substring fallback continues to
work, so this commit is safe to ship independently.
OpenAI-compatible LLM gateway. Pairs with the librefang-llm-drivers
registration (librefang PR #3076) so Novita models surface in the
dashboard model picker without each user having to add them via
/api/models/custom.
Pricing and limits sourced from GET /openai/v1/models on 2026-04-25
(input_token_price_per_m / output_token_price_per_m, divided by 10000
to get USD per million tokens). Curated to a small popular subset:
- deepseek/deepseek-v3.2 (frontier)
- moonshotai/kimi-k2-thinking (frontier, supports_thinking)
- minimax/minimax-m2 (smart)
- meta-llama/llama-3.3-70b-instruct (smart)
- qwen/qwen3-coder-30b-a3b-instruct (balanced)
- zai-org/glm-4.7-flash (balanced)
Evan