fix(vision): guard image pixel dimensions, not just bytes (#37677)

Anthropic enforces two independent ceilings per image:
1. 5 MB encoded byte size
2. 8000 px longest side

Hermes only guarded #1. A tall screenshot (e.g. 1200x12000 at 0.06 MB)
passes every byte check but fails the pixel check, returning a
non-retryable HTTP 400 that permanently bricks the conversation thread.

Fixes:
- error_classifier: add 'image dimensions exceed' pattern to
  _IMAGE_TOO_LARGE_PATTERNS so the 400 is classified as image_too_large
  and triggers the shrink/retry path instead of falling through to
  non-retryable error.
- conversation_compression: check pixel dimensions (via Pillow) even
  when byte size is under the 4 MB target. If max(dims) > 8000, force
  shrink.
- vision_tools._resize_image_for_vision: add optional max_dimension param.
  When set, images exceeding the pixel cap are downscaled even if they're
  under the byte budget. The resize loop now checks both byte AND pixel
  limits before accepting a candidate.

Closes #37677
This commit is contained in:
kyssta-exe 2026-06-02 23:45:22 +00:00 committed by Teknium
parent f7dabd3019
commit 6bdbe30763
4 changed files with 65 additions and 10 deletions

View file

@ -171,6 +171,7 @@ _IMAGE_TOO_LARGE_PATTERNS = [
"image too large", # generic
"image_too_large", # error_code variant
"image size exceeds", # variant
"image dimensions exceed", # Anthropic: "image dimensions exceed max allowed size: 8000 pixels"
# "request_too_large" on a request known to contain an image → image is
# the likely culprit; we still try the shrink path before giving up.
]