fix(models): read OpenRouter live context_length before hardcoded catch-all (#42986)

OpenRouter-routed slugs that are absent from models.dev (e.g. a freshly
shipped anthropic/claude-fable-5) fell through to the generic
DEFAULT_CONTEXT_LENGTHS["claude"]=200K entry and under-reported their real
1M window. The step-6 OpenRouter live-metadata fallback was gated on
`not effective_provider`, but an OpenRouter selection sets
effective_provider="openrouter" (inferred from the base URL), so that
branch was dead code for every OR model.

Add a dedicated step-5 OpenRouter branch that consults the live /models
catalog (authoritative, refreshes as new slugs ship) before models.dev and
the hardcoded family defaults — mirroring the existing Nous/Copilot/GMI
branches. Keeps the Kimi-family 32k underreport guard. Per-model values are
respected (claude-haiku-4.5 stays 200K), so it does not blanket-bump to 1M.

Regression tests cover the fable-5 case, the genuinely-200k case, and the
Kimi guard.
This commit is contained in:
Teknium
2026-06-09 10:49:32 -07:00
committed by GitHub
parent f6f573ebaa
commit 967c325da8
2 changed files with 75 additions and 0 deletions
+22
View File
@@ -1810,6 +1810,28 @@ def get_model_context_length(
if ctx is not None:
save_context_length(model, base_url, ctx)
return ctx
# 5f. OpenRouter live /models metadata — authoritative for OpenRouter-routed
# models. OpenRouter's catalog carries per-model context_length (e.g.
# anthropic/claude-fable-5 -> 1M) and refreshes as new slugs ship, so it
# must win over both models.dev (step 5g) and the hardcoded family catch-all
# (step 8). Before this branch, an OpenRouter selection set
# effective_provider="openrouter", which (a) made the models.dev lookup miss
# brand-new slugs and (b) skipped the step-6 OR fallback (gated on `not
# effective_provider`), so a fresh slug like claude-fable-5 fell through to
# the generic "claude": 200K entry and under-reported a 1M window. Mirrors
# the dedicated Nous/Copilot/GMI branches above.
if effective_provider == "openrouter":
metadata = fetch_model_metadata()
entry = metadata.get(model)
if entry:
or_ctx = entry.get("context_length")
# Guard against the known OpenRouter Kimi-family 32k underreport
# (same class the hardcoded overrides exist to mitigate).
if isinstance(or_ctx, int) and or_ctx > 0 and not (
or_ctx == 32768 and _model_name_suggests_kimi(model)
):
return or_ctx
if effective_provider:
from agent.models_dev import lookup_models_dev_context
ctx = lookup_models_dev_context(effective_provider, model)