Skip to content

EthenEthenEthen

LoginPlatformTry Ethen

Open Source Model Profile · open-thoughts

OpenThinker-7B

OpenThinker-7B is a 7.62B-parameter Qwen2 reasoning fine-tune from open-thoughts. Its card documents Qwen2.5-7B-Instruct training on DeepSeek-R1-distilled OpenThoughts-114k.

Publisher
open-thoughts
Task
text-generation
Model type
qwen2
License
apache-2.0
Library
transformers
Publication status
Accepted · not indexed

Model overview

OpenThinker-7B is published by open-thoughts as a Qwen2 text-generation fine-tune. The captured configuration identifies Qwen2ForCausalLM and Safetensors metadata reports 7,615,616,512 parameters, or about 7.62B. The model card describes it as Qwen/Qwen2.5-7B-Instruct fine-tuned on OpenThoughts-114k distilled from DeepSeek-R1.

Recorded capabilities

OpenThoughts-114k reasoning distill

The model card describes fine-tuning Qwen/Qwen2.5-7B-Instruct on OpenThoughts-114k, a dataset derived by distilling DeepSeek-R1.

Publisher-reported reasoning scores

According to the model card, OpenThinker-7B reports AIME24 31.3, MATH500 83.0, GPQA-Diamond 42.4, and LCBv2 All 39.9, above the card's Bespoke-Stratos-7B row.

Documented multi-node training run

The model card states training on four 8xH100 nodes for 20 hours over 3.0 epochs, with learning rate 1e-05, cosine scheduling, and Transformers 4.46.1.

Use cases in the source record

  • Reasoning-oriented text-generation experiments using the documented DeepSeek-R1-distilled OpenThoughts-114k lineage.
  • Benchmark-aware comparisons on AIME, MATH500, GPQA-Diamond, and LiveCodeBench-style tasks, reusing the publisher's reported figures as baselines.

Limitations and unknowns

  • Provider state is historical snapshot data and should be refreshed before being presented as current.
  • No context-window value was extracted from this record.
  • No VRAM, quantization, pricing, or current deployment figures were extracted.
  • Benchmark figures are publisher-reported comparisons, not Ethen-measured results, and should not be generalized into universal superiority.

Source and provenance

Source: open-thoughts/OpenThinker-7B

Captured: Unknown. Processed: 2026-09-07T19:35:27.848403+00:00.

We have released a paper for OpenThoughts! See our paper here . OpenThinker-7B This model is a fine-tuned version of Qwen/Qwen2.5-7B-Instruct on the OpenThoughts-114k dataset dataset. The dataset is derived by distilling DeepSeek-R1 using the data pipeline available on github . More info about the dataset can be found on the dataset card at OpenThoughts-114k dataset . This model improves upon the Bespoke-Stratos-7B model , which used 17k examples ( Bespoke-Stratos-17k dataset ). The numbers reported in the table below are evaluated with our open-source tool Evalchemy . AIME24 MATH500 GPQA-Diamond LCBv2 Easy LCBv2 Medium LCBv2 Hard L…

F001F002F003F004F005F006F007F009F010F012F013F016F017F018