feat: OpenRouter support + 7 new AIMO3 hard problems#1
Open
sonpham-org wants to merge 2 commits intomasterfrom
Open
feat: OpenRouter support + 7 new AIMO3 hard problems#1sonpham-org wants to merge 2 commits intomasterfrom
sonpham-org wants to merge 2 commits intomasterfrom
Conversation
added 2 commits
March 29, 2026 00:54
- Add openrouter:// URL scheme to create_client() - LLMClient now accepts api_key (Bearer auth header) and model_name (required by OpenRouter for model routing) - Fix api key resolution for non-Gemini remote APIs (checks OPENROUTER_API_KEY env when model URL is openrouter://) - Fix ping() to skip /health check for remote APIs - Update models.json with 4 OpenRouter free models: - openai/gpt-oss-120b:free (same model as calibration data) - openai/gpt-oss-20b:free - nvidia/nemotron-3-super-120b-a12b:free - qwen/qwen3-next-80b-a3b-instruct:free - Add 7 new AIMO3 hard problems (all pass_rate=0.125 on 20B, diverse topics, high solution token count): aimo3_hard_5f9595ae, aimo3_hard_f728b4b1, aimo3_hard_b0dc264b, aimo3_hard_858cba58, aimo3_hard_00eaa992, aimo3_hard_b1da52fa, aimo3_hard_2e0b7ba3 Tested: gpt-oss-20b on digit_sum_ten → minimum_window=672 ✓
- Increase backoff retries 5→20, max_delay 60s→120s - Respect Retry-After / x-ratelimit-reset-requests headers - Add X-OpenRouter-Cache: true header for prompt caching on repeated system prompts - Drop parallelism to max 3 per model with staggered launches to avoid 429 bursts
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Changes
OpenRouter backend support
openrouter://URL scheme tocreate_client()LLMClientnow acceptsapi_key(Bearer auth header) andmodel_name(required by OpenRouter)OPENROUTER_API_KEYenv var when URL scheme isopenrouter://ping()skips/healthcheck for remote APIsmodels.json — 4 new free OpenRouter models
openai/gpt-oss-120b:free(same model as calibration data)openai/gpt-oss-20b:freenvidia/nemotron-3-super-120b-a12b:free(262K ctx)qwen/qwen3-next-80b-a3b-instruct:free(262K ctx)7 new AIMO3 hard problems
All pass_rate=0.125 on GPT-OSS-20B, diverse topics:
aimo3_hard_5f9595ae, aimo3_hard_f728b4b1, aimo3_hard_b0dc264b, aimo3_hard_858cba58, aimo3_hard_00eaa992, aimo3_hard_b1da52fa, aimo3_hard_2e0b7ba3
Tested
gpt-oss-20b on digit_sum_ten → minimum_window=672 ✓