Skywork-o1
China's first open o1-style reasoning model, built on Llama 3.1 8B
About
Skywork-o1 was Kunlun's late-2024 entry in the race to replicate o1-style reasoning: the open variant, Skywork-o1-Open, builds on Llama 3.1 8B with slow-thinking capabilities for math, coding, and logical reasoning, and invitation testing began November 27, 2024. The team reported meaningful gains on math and coding benchmarks over the base model. Kunlun later followed with Skywork-OR1 in May 2025, a full reinforcement-learning-trained reasoner with open weights, training code, and datasets. The caveat is coverage: most documentation is Chinese and independent Western evaluation is thin, so benchmark claims should be treated cautiously until third parties reproduce them.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Team-reported: improved math and coding scores over the Llama 3.1 8B base (team-reported; thin independent verification)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Announced Nov 2024, invitation testing from 2024-11-27 per Chinese press; Open variant built on Llama 3.1 8B with o1-style reasoning; Benchmark gains are team-reported with thin independent verification — flagged; Mostly Chinese documentation; exact weight-release day not pinned — releasedAt null; Skywork-OR1 (May 2025) follow-up with open weights/code/data verified via arXiv.