LLMs TigerLive LLM benchmarks

seed-oss-36b-instruct: benchmark scores, price and ranking

seed-oss-36b-instruct from ByteDance-Seed ranks #133 of 743 LLMs on the LLMs Tiger aggregate benchmark score (51.3), based on 4 benchmark results captured up to 2026-10-13.

Organisation
ByteDance-Seed
Type
Open-weight / local model
Open weights
Yes
Released
2025-08-20 (source: epoch)
Aggregate rank
#133 of 743 (score 51.3)

Relative to other models, seed-oss-36b-instruct is strongest on MMLU-Pro (#24 of 136) and weakest on GPQA Diamond (#139 of 275).

Across the 4 benchmarks where at least ten models are scored, seed-oss-36b-instruct is in the top 10% on 0 and in the top half on 3.

seed-oss-36b-instruct benchmark results

seed-oss-36b-instruct benchmark scores, ranked against all models measured on each benchmark
BenchmarkScoreRankSourceCaptured
MMLU-Pro82.7%#24 of 136HuggingFace MMLU-Pro leaderboard2026-10-13
OTIS Mock AIME67.5%#83 of 189Epoch AI Benchmarking Hub2026-10-13
Chess Puzzles13.0%#61 of 122Epoch AI Benchmarking Hub2026-10-13
GPQA Diamond71.5%#139 of 275Epoch AI Benchmarking Hub2026-10-13

About seed-oss-36b-instruct

seed-oss-36b-instruct is an open-weight model that can run on your own hardware from ByteDance-Seed. It was released on 2025-08-20, according to Epoch AI. No API price is listed for it yet. We hold 4 benchmark results for it across 2 areas, and none is in the top tenth of its board.

Where seed-oss-36b-instruct stands

On our aggregate score, seed-oss-36b-instruct is #133 of 743 ranked models, in the top quarter. Its score is 51.3; the scale is centred on 50, so it sits 1.3 points above the typical model. That beats 82% of the ranked field. Among the 3 ranked ByteDance-Seed models it is #1. The next one down is Seed-OSS-36B-Base at 49.2.

seed-oss-36b-instruct, benchmark by benchmark

MMLU-Pro puts seed-oss-36b-instruct at 82.7%, #24 of 136. That beats 82% of the models tested, but 5.4 points short of Darwin-180B-RSI (88.1%). The results either side are LongCat-Flash-Chat (82.7%) and MiniMax-M2 (82.0%). It covers 57-subject knowledge + reasoning. The board median is 55.3%, so seed-oss-36b-instruct is 27.4 points above it. Source: HuggingFace MMLU-Pro leaderboard, captured 2026-10-13.

On OTIS Mock AIME, seed-oss-36b-instruct reaches 67.5%. It ranks #83 of 189, ahead of 56% of the field, and 32.5 points behind Claude Fable 5.1 (max) at 100.0%. The results either side are GPT-5.5 Instant (68.1%) and Qwen3-32B (66.9%). It covers competition maths (AIME 2024-25 style). The board median is 57.8%, so seed-oss-36b-instruct is 9.7 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

seed-oss-36b-instruct manages 13.0% on Chess Puzzles, good for #61 of 122. GPT-6 Astra (max) leads at 72.0%, 59.0 points further up, and 50% of models sit below. The results either side are Qwen3.5 397B-A17B (13.0%) and GPT-5.5 Instant (12.0%). It covers chess tactics puzzles. The board median is 12.5%, so seed-oss-36b-instruct is 0.5 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

seed-oss-36b-instruct lands below the middle on GPQA Diamond: 71.5%, #139 of 275, with 50% of the field scoring higher. GPT-6 Astra (max) leads at 95.8%; the closest result above is GLM-4.5-Air at 71.7%. It covers expert-level science questions. The board median is 71.7%, so seed-oss-36b-instruct is 0.2 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Strengths and weaknesses

Reasoning, maths and science: solid, ahead of 63% of the field on average across 3 benchmarks. Its best board here is MMLU-Pro (#24 of 136) and its weakest is GPQA Diamond (#139 of 275).

Reading these numbers

The 4 scores on this page come from 2 sources: Epoch AI Benchmarking Hub (3), HuggingFace MMLU-Pro leaderboard (1). No official leaderboard supplied any of them, so treat them as third-party measurements. With 4 results the aggregate rank can move a lot when one new score arrives.

Models ranked nearby

More from ByteDance-Seed

Compare it on the live leaderboard → · How the aggregate score works