Spaces:
Running
Running
File size: 14,652 Bytes
261c237 e99df76 2415446 e99df76 2415446 691dc7c 8e438ab 2e7b8ca 8e438ab e99df76 61bb677 2415446 e99df76 a9e27c0 691dc7c 2e7b8ca 8e438ab 61bb677 e99df76 61bb677 2e7b8ca a9e27c0 2e7b8ca 61bb677 e99df76 a9e27c0 61bb677 e99df76 a9e27c0 61bb677 abe1c43 a9e27c0 61bb677 2415446 3934d27 2415446 f7872b8 3934d27 f7872b8 61bb677 2415446 3934d27 61bb677 2415446 e99df76 8e438ab 2e7b8ca a9e27c0 2415446 e99df76 3934d27 2415446 3934d27 2415446 61bb677 2415446 61bb677 2415446 61bb677 2415446 61bb677 2415446 3934d27 2415446 3934d27 61bb677 2415446 61bb677 3934d27 2415446 3934d27 2415446 3934d27 2415446 3934d27 f7872b8 3934d27 f7872b8 3934d27 f7872b8 3934d27 2415446 3934d27 f7872b8 3934d27 f7872b8 3934d27 f7872b8 2415446 691dc7c 2415446 61bb677 691dc7c 8e438ab 2e7b8ca 2415446 691dc7c 61bb677 2415446 61bb677 2415446 61bb677 691dc7c 61bb677 2415446 61bb677 2415446 61bb677 2415446 61bb677 2415446 61bb677 2415446 e99df76 f7872b8 e99df76 f7872b8 e99df76 2415446 3934d27 61bb677 2415446 8e438ab 2415446 e99df76 2415446 e99df76 2415446 3934d27 2415446 8e438ab 2e7b8ca a9e27c0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 | from __future__ import annotations
"""Model routing for Claude-compatible requests."""
from dataclasses import dataclass, replace
from loguru import logger
from free_claude_code.application.errors import (
InvalidRequestError,
UnknownProviderError,
)
from free_claude_code.config.model_refs import (
is_retired_model_ref,
parse_model_name,
parse_provider_type,
replace_retired_model_ref,
)
from free_claude_code.config.provider_catalog import PROVIDER_CATALOG
from free_claude_code.config.reasoning import ReasoningPreference
from free_claude_code.config.settings import Settings
from free_claude_code.core.anthropic import (
MessagesRequest,
NativeTokenCountRequest,
TokenCountRequest,
)
from free_claude_code.core.anthropic.passthrough import NativeMessagesRequest
from free_claude_code.core.chat_completions.request import ChatRequest
from free_claude_code.core.gateway_model_ids import (
DESKTOP_MODEL_PREFIX,
DESKTOP_NO_THINKING_PREFIX,
decode_gateway_model_id,
)
from free_claude_code.core.gemini import GeminiRequest
from free_claude_code.core.openai_responses import OpenAIResponsesRequest
from free_claude_code.core.reasoning import ReasoningPolicy
from free_claude_code.core.trace import trace_event
from .reasoning import (
resolve_chat_reasoning_policy,
resolve_gemini_reasoning_policy,
resolve_reasoning_policy,
resolve_responses_reasoning_policy,
)
_ROUTE_SETTINGS = (
("fable", "model_fable", "reasoning_fable"),
("opus", "model_opus", "reasoning_opus"),
("sol", "model_opus", "reasoning_opus"),
("gpt-5.6-sol", "model_opus", "reasoning_opus"),
("haiku", "model_haiku", "reasoning_haiku"),
("luna", "model_haiku", "reasoning_haiku"),
("gpt-5.6-luna", "model_haiku", "reasoning_haiku"),
("sonnet", "model_sonnet", "reasoning_sonnet"),
("terra", "model_sonnet", "reasoning_sonnet"),
("gpt-5.6-terra", "model_sonnet", "reasoning_sonnet"),
)
@dataclass(frozen=True, slots=True)
class ProviderModelTarget:
"""One canonical provider/model execution target."""
provider_id: str
provider_model: str
provider_model_ref: str
api_key_number: int = 1
@dataclass(frozen=True, slots=True)
class ProviderModelCandidate:
"""One execution target with a candidate-local reasoning preference."""
target: ProviderModelTarget
reasoning_preference: ReasoningPreference = ReasoningPreference.INHERIT
@dataclass(frozen=True, slots=True)
class ResolvedModelRoute:
"""One public model resolved to a primary and ordered fallbacks."""
original_model: str
primary: ProviderModelTarget
fallbacks: tuple[ProviderModelCandidate, ...]
reasoning_preference: ReasoningPreference
@dataclass(frozen=True, slots=True)
class RoutedMessagesRequest:
request: MessagesRequest
resolved: ResolvedModelRoute
reasoning: ReasoningPolicy
@dataclass(frozen=True, slots=True)
class RoutedNativeMessagesRequest:
request: NativeMessagesRequest
resolved: ResolvedModelRoute
def supports_native_messages(provider_id: str) -> bool:
descriptor = PROVIDER_CATALOG.get(provider_id)
return descriptor is not None and descriptor.native_messages_passthrough
@dataclass(frozen=True, slots=True)
class RoutedResponsesRequest:
request: OpenAIResponsesRequest
resolved: ResolvedModelRoute
reasoning: ReasoningPolicy
@dataclass(frozen=True, slots=True)
class RoutedGeminiRequest:
request: GeminiRequest
resolved: ResolvedModelRoute
reasoning: ReasoningPolicy
@dataclass(frozen=True, slots=True)
class RoutedChatRequest:
request: ChatRequest
resolved: ResolvedModelRoute
reasoning: ReasoningPolicy
@dataclass(frozen=True, slots=True)
class RoutedTokenCountRequest:
request: TokenCountRequest | NativeTokenCountRequest
resolved: ResolvedModelRoute
class ModelRouter:
"""Resolve incoming Claude model names to configured provider/model pairs."""
def __init__(self, settings: Settings):
self._settings = settings
def resolve(self, claude_model_name: str) -> ResolvedModelRoute:
(
direct_provider_id,
direct_provider_model,
force_reasoning_off,
) = self._direct_provider_model(claude_model_name)
if direct_provider_id is not None and direct_provider_model is not None:
reasoning_preference = (
ReasoningPreference.OFF
if force_reasoning_off
else self._settings.reasoning_policy
)
logger.debug(
"MODEL DIRECT: '{}' -> provider='{}' model='{}' reasoning={}",
claude_model_name,
direct_provider_id,
direct_provider_model,
reasoning_preference.value,
)
primary = self._target(direct_provider_id, direct_provider_model)
return ResolvedModelRoute(
original_model=claude_model_name,
primary=primary,
fallbacks=self._fallback_targets(primary),
reasoning_preference=reasoning_preference,
)
provider_model_ref = self._resolve_model_ref(claude_model_name)
reasoning_preference = self._resolve_reasoning_preference(claude_model_name)
primary = self._target_from_ref(provider_model_ref)
if primary.provider_model != claude_model_name:
logger.debug(
"MODEL MAPPING: '{}' -> '{}'",
claude_model_name,
primary.provider_model,
)
return ResolvedModelRoute(
original_model=claude_model_name,
primary=primary,
fallbacks=self._fallback_targets(primary),
reasoning_preference=reasoning_preference,
)
def _target_from_ref(
self, provider_model_ref: str, api_key_number: int = 1
) -> ProviderModelTarget:
return self._target(
parse_provider_type(provider_model_ref),
parse_model_name(provider_model_ref),
api_key_number,
)
def _target(
self, provider_id: str, provider_model: str, api_key_number: int = 1
) -> ProviderModelTarget:
self._validate_provider_id(provider_id)
return ProviderModelTarget(
provider_id=provider_id,
provider_model=provider_model,
provider_model_ref=f"{provider_id}/{provider_model}",
api_key_number=api_key_number,
)
def _fallback_targets(
self, primary: ProviderModelTarget
) -> tuple[ProviderModelCandidate, ...]:
configured = tuple(
ProviderModelCandidate(
self._target_from_ref(entry.model, entry.api_key_number),
entry.reasoning,
)
for entry in self._settings.model_fallbacks or ()
)
return tuple(
candidate for candidate in configured if candidate.target != primary
)
def _validate_provider_id(self, provider_id: str) -> None:
if provider_id not in self._settings.provider_ids:
raise UnknownProviderError.for_provider(
provider_id, self._settings.provider_ids
)
def _direct_provider_model(
self, model_name: str
) -> tuple[str | None, str | None, bool]:
try:
decoded = decode_gateway_model_id(model_name)
except ValueError:
raise InvalidRequestError("Invalid Claude Desktop model ID") from None
candidate = (
f"{decoded.provider_id}/{decoded.provider_model}"
if decoded is not None
else model_name
)
if is_retired_model_ref(candidate):
return (
parse_provider_type(self._settings.model),
parse_model_name(self._settings.model),
decoded.force_reasoning_off if decoded is not None else False,
)
replacement = replace_retired_model_ref(candidate)
if replacement != candidate:
return (
parse_provider_type(replacement),
parse_model_name(replacement),
decoded.force_reasoning_off if decoded is not None else False,
)
if decoded is not None:
if decoded.provider_id not in self._settings.provider_ids:
if decoded.provider_id.startswith("custom_") or model_name.partition(
"/"
)[0] in {
DESKTOP_MODEL_PREFIX,
DESKTOP_NO_THINKING_PREFIX,
}:
raise UnknownProviderError.for_provider(
decoded.provider_id, self._settings.provider_ids
)
return None, None, False
return (
decoded.provider_id,
decoded.provider_model,
decoded.force_reasoning_off,
)
provider_id, separator, provider_model = model_name.partition("/")
if not separator:
return None, None, False
if provider_id not in self._settings.provider_ids:
if provider_id.startswith("custom_"):
raise UnknownProviderError.for_provider(
provider_id, self._settings.provider_ids
)
return None, None, False
if not provider_model:
return None, None, False
return provider_id, provider_model, False
def _resolve_model_ref(self, claude_model_name: str) -> str:
"""Resolve a Claude model name to the configured provider/model ref."""
route = self._matched_route(claude_model_name)
if route is not None:
model = getattr(self._settings, route[1])
if isinstance(model, str):
return model
return self._settings.model
def _resolve_reasoning_preference(
self, claude_model_name: str
) -> ReasoningPreference:
"""Resolve a route override without inspecting the provider model."""
route = self._matched_route(claude_model_name)
if route is not None:
preference = getattr(self._settings, route[2])
if preference is not ReasoningPreference.INHERIT:
return preference
return self._settings.reasoning_policy
@staticmethod
def _matched_route(model_name: str) -> tuple[str, str, str] | None:
normalized = model_name.lower()
return next(
(route for route in _ROUTE_SETTINGS if route[0] in normalized),
None,
)
def route_native_messages(
self,
request: NativeMessagesRequest,
resolved: ResolvedModelRoute,
) -> RoutedNativeMessagesRequest:
eligible = tuple(
candidate
for candidate in resolved.fallbacks
if supports_native_messages(candidate.target.provider_id)
)
trace_event(
stage="routing",
event="free_claude_code.api.route.messages_contract",
source="application",
contract="native",
excluded_fallbacks=tuple(
candidate.target.provider_model_ref
for candidate in resolved.fallbacks
if candidate not in eligible
),
)
return RoutedNativeMessagesRequest(
request, replace(resolved, fallbacks=eligible)
)
def resolve_messages_request(
self, request: MessagesRequest
) -> RoutedMessagesRequest:
"""Return an internal routed request context."""
resolved = self.resolve(request.model)
routed = request.model_copy(deep=True)
routed.model = resolved.primary.provider_model
return RoutedMessagesRequest(
request=routed,
resolved=resolved,
reasoning=resolve_reasoning_policy(
routed,
resolved.reasoning_preference,
),
)
def resolve_messages_request_with_policy(
self,
request: MessagesRequest,
*,
reasoning: ReasoningPolicy,
) -> RoutedMessagesRequest:
"""Route an application-owned Messages request with resolved intent."""
resolved = self.resolve(request.model)
routed = request.model_copy(
update={"model": resolved.primary.provider_model},
deep=True,
)
return RoutedMessagesRequest(
request=routed,
resolved=resolved,
reasoning=reasoning,
)
def resolve_token_count_request(
self,
request: TokenCountRequest | NativeTokenCountRequest,
*,
resolved: ResolvedModelRoute | None = None,
) -> RoutedTokenCountRequest:
"""Return an internal token-count request context."""
resolved = resolved or self.resolve(request.model)
routed = request.model_copy(
update={"model": resolved.primary.provider_model}, deep=True
)
return RoutedTokenCountRequest(request=routed, resolved=resolved)
def resolve_responses_request(
self,
request: OpenAIResponsesRequest,
) -> RoutedResponsesRequest:
"""Route a Responses request without converting its protocol shape."""
resolved = self.resolve(request.model)
routed = request.model_copy(deep=True)
routed.model = resolved.primary.provider_model
return RoutedResponsesRequest(
request=routed,
resolved=resolved,
reasoning=resolve_responses_reasoning_policy(
routed,
resolved.reasoning_preference,
),
)
def resolve_gemini_request(self, request: GeminiRequest) -> RoutedGeminiRequest:
resolved = self.resolve(request.model)
return RoutedGeminiRequest(
request=request.model_copy(
update={"model": resolved.primary.provider_model}, deep=True
),
resolved=resolved,
reasoning=resolve_gemini_reasoning_policy(
request, resolved.reasoning_preference
),
)
def resolve_chat_request(self, request: ChatRequest) -> RoutedChatRequest:
resolved = self.resolve(request.model)
return RoutedChatRequest(
request=request.model_copy(
update={"model": resolved.primary.provider_model}, deep=True
),
resolved=resolved,
reasoning=resolve_chat_reasoning_policy(
request, resolved.reasoning_preference
),
)
|