Evan
6785807633
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.
136 lines
3.4 KiB
TOML
136 lines
3.4 KiB
TOML
# BytePlus ModelArk — Coding Plan (Anthropic-compatible) — https://www.byteplus.com/en/product/ModelArk
|
|
# International edition's Claude Code-compatible endpoint with friendly model aliases.
|
|
# Models: 9 (auto-routed alias + 8 named models)
|
|
#
|
|
# 💡 BILLING — calls here count against your **Coding Plan
|
|
# subscription quota**, not your per-token ModelArk USD balance. If you
|
|
# don't have a Coding Plan or want per-token billing instead, use the
|
|
# `byteplus` provider in byteplus.toml. The Coding Plan endpoint also
|
|
# does not expose image / video / embedding models — only the text
|
|
# models listed below.
|
|
#
|
|
# Pricing fields below mirror the underlying versioned snapshots in
|
|
# byteplus.toml for catalog accounting purposes; actual subscription
|
|
# consumption is tracked by the BytePlus console, not these numbers.
|
|
#
|
|
# The Coding Plan endpoint exposes friendly stable names that route to
|
|
# the current production snapshot, so versions advance silently without
|
|
# config changes (use `ark-code-latest` to always get the best current
|
|
# routing).
|
|
|
|
[provider]
|
|
id = "byteplus_coding"
|
|
display_name = "BytePlus Coding Plan"
|
|
api_key_env = "BYTEPLUS_CODING_API_KEY"
|
|
base_url = "https://ark.ap-southeast.bytepluses.com/api/coding"
|
|
key_required = true
|
|
|
|
[[models]]
|
|
id = "ark-code-latest"
|
|
display_name = "Ark Code (auto-routed)"
|
|
tier = "frontier"
|
|
context_window = 196608
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 1.00
|
|
output_cost_per_m = 6.00
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
aliases = ["ark-coding"]
|
|
|
|
[[models]]
|
|
id = "bytedance-seed-code"
|
|
display_name = "ByteDance Seed Code"
|
|
tier = "frontier"
|
|
context_window = 196608
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 1.00
|
|
output_cost_per_m = 6.00
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
aliases = []
|
|
|
|
[[models]]
|
|
id = "dola-seed-2.0-pro"
|
|
display_name = "Dola Seed 2.0 Pro"
|
|
tier = "frontier"
|
|
context_window = 196608
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 1.00
|
|
output_cost_per_m = 6.00
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
aliases = []
|
|
|
|
[[models]]
|
|
id = "dola-seed-2.0-code"
|
|
display_name = "Dola Seed 2.0 Code"
|
|
tier = "smart"
|
|
context_window = 196608
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 1.00
|
|
output_cost_per_m = 6.00
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
aliases = []
|
|
|
|
[[models]]
|
|
id = "dola-seed-2.0-lite"
|
|
display_name = "Dola Seed 2.0 Lite"
|
|
tier = "smart"
|
|
context_window = 196608
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 0.50
|
|
output_cost_per_m = 4.00
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
aliases = []
|
|
|
|
[[models]]
|
|
id = "kimi-k2.5"
|
|
display_name = "Kimi K2.5"
|
|
tier = "frontier"
|
|
context_window = 196608
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 0.44
|
|
output_cost_per_m = 2.0
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
reasoning_echo_policy = "empty_string"
|
|
aliases = []
|
|
|
|
[[models]]
|
|
id = "glm-4.7"
|
|
display_name = "GLM-4.7"
|
|
tier = "smart"
|
|
context_window = 196608
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 0.38
|
|
output_cost_per_m = 1.74
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
aliases = []
|
|
|
|
[[models]]
|
|
id = "glm-5.1"
|
|
display_name = "GLM-5.1"
|
|
tier = "frontier"
|
|
context_window = 196608
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 1.05
|
|
output_cost_per_m = 3.5
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
aliases = []
|
|
|
|
[[models]]
|
|
id = "gpt-oss-120b"
|
|
display_name = "GPT-OSS 120B"
|
|
tier = "smart"
|
|
context_window = 131072
|
|
max_output_tokens = 8192
|
|
input_cost_per_m = 0.039
|
|
output_cost_per_m = 0.18
|
|
supports_tools = true
|
|
supports_streaming = true
|
|
aliases = []
|