Seed-OSS-36B
Apache-2.0 open 36B model with 512K context for reasoning and agents.
About
ByteDance's Seed team released Seed-OSS-36B as its first open model family: 36B parameters across base, base-without-synthetic-data, and instruct variants, all under Apache 2.0 with no licensing fees for commercial use. The standout spec is a native 512K-token context window, aimed at long documents and extended reasoning chains, and the team says it trains well on only 12T tokens.
It is explicitly positioned for international (non-China) use cases. As with most vendor-reported results, treat the benchmark claims as directional until independent evals land.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
- Native 512K-token context window
- Trained on 12T tokens; strong results on open benchmarks (ByteDance-reported)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: ByteDance Seed team official repo + HF org, Apache-2.0 with no commercial restriction; 512K context and 12T-token training figure from official model card; Benchmark claims are vendor-reported; independent evals not yet confirmed. License: Apache-2.0 confirmed from official GitHub repo and HF model card..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 9.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · ByteDance Seed team official repo + HF org, Apache-2.0 with no commercial restriction
- · 512K context and 12T-token training figure from official model card
- · Benchmark claims are vendor-reported; independent evals not yet confirmed