UC Berkeley RAIL et al.
Octo
93M Commercial use OK 8/8 verified
Facts
- Released
- 2024-05 1 Source
- Parameters
- 0.093 B 1 Source
- Architecture
- Transformer policy 0 Source
- License
- MIT (tagged on code and weights repositories) 2 Source
- Commercial
- Commercial use OK 0 Source
- Weights
- huggingface.co ↗ 1 Source
- Code
- github.com ↗ 1 Source
- Paper
- arxiv.org ↗ 0 Source
- Input modalities
- rgb · language · goal_image 0 Source
- Action space
- End-effector pose delta + gripper (diffusion) 0 Source
- Cross-embodiment
- Pretrained on approx. 800k Open X-Embodiment trajectories 1 Source
- Pretraining data (hours)
- Unverified 0 Source
- Target forms
- Single arm 0 Source
Architecture · Transformer backbone + diffusion action head; octo-base 93M, octo-small 27M
Sources · 2 sources
params_b,license,code_url,cross_embodiment https://github.com/octo-models/octo Official page · Fetched 2026-09-04 · No snapshot stored yet README：Octo-Base 93M Params / Octo-Small 27M；License: MIT；800k robot trajectories from Open X-Embodiment
weights_url,license,release_date https://huggingface.co/api/models/rail-berkeley/octo-base-1.5 Model hub page · Fetched 2026-09-04 · No snapshot stored yet Model card license: mit; createdAt 2024-05-21
Activity signals
Public counters fetched daily from GitHub, Hugging Face and ModelScope. Numbers only; no judgement. · fetched 2026-10-18
GitHub stars
1,807
+28 / 30 d
Commits · 90 d
0
last push 810 d ago
HF downloads · 30 d
208
all-time 3,147 · 18 likes
Weights last modified
2024-05-21
881 d ago
GitHubocto-models/octo · 282 forks · 97 open issues · latest release v1.5 (2024-05-24)no push for 809 d
ModelScopeno same-name mirror found (searched by repo name)
Hugging Facerail-berkeley/octo-base-1.5
stars · last 30 d
Public benchmark scores
As printed in the source; protocols differ between papers. Full table with notes on the benchmarks page.
| Benchmark | Subset | Score | Source |
|---|---|---|---|
| LIBEROOcto-Base | avg | 75.1 | third-party · OpenVLA paper, Table 12, row 'Octo fine-tuned', Average (75.1 ± 0.6%) |
| LIBEROOcto-Base | goal | 84.6 | third-party · OpenVLA paper, Table 12, row 'Octo fine-tuned', LIBERO-Goal (84.6 ± 0.9%) |
| LIBEROOcto-Base | long | 51.1 | third-party · OpenVLA paper, Table 12, row 'Octo fine-tuned', LIBERO-Long (51.1 ± 1.3%) |
| LIBEROOcto-Base | object | 85.7 | third-party · OpenVLA paper, Table 12, row 'Octo fine-tuned', LIBERO-Object (85.7 ± 0.9%) |
| LIBEROOcto-Base | spatial | 78.9 | third-party · OpenVLA paper, Table 12, row 'Octo fine-tuned', LIBERO-Spatial (78.9 ± 1.0%) |
| Real-robot evaluation (self-reported)Octo-Base | finetuning_6_domains_avg | 72 | paper · Table I (fine-tuning results), row 'Octo', Average |
| SimplerEnv · Google RobotOcto-Base | variant_aggregation | 1.2 | third-party · CogACT paper, Table 1 (Variant Aggregation), row 'Octo-Base', Average |
| SimplerEnv · Google RobotOcto-Base | visual_matching | 11 | third-party · CogACT paper, Table 1 (Visual Matching), row 'Octo-Base', Average |
| SimplerEnv · WidowX (Bridge)Octo-Base | visual_matching | 17.5 | third-party · CogACT paper, Table 2 (WidowX), row 'Octo-Base', Average |
| SimplerEnv · WidowX (Bridge)Octo-Small | visual_matching | 26.7 | third-party · CogACT paper, Table 2 (WidowX), row 'Octo-Small', Average |
Deployment recipe
No recipe yet. We publish only steps that actually ran on a rented GPU with a measured latency; this model has not been through that yet.
Record history
2026-09-04
Sources fetched
params_b,license,code_url,cross_embodiment · weights_url,license,release_date
Latency & cost
cost = p50 ms ÷ 1000 × rental ÷ 3600 × 1000
| Hardware | p50 ms | p95 ms | Peak VRAM | Cost per 1k inferences ¥ | samples |
|---|---|---|---|---|---|
NVIDIA RTX 4090 ¥1.68/h · GongJi Compute (2026) | — | — | — | — no latency data | No data |
NVIDIA RTX 5090 ¥2.4/h · Runyun 2.4 / GongJi 2.5 / Paratera Cloud 2.98 (2026) | — | — | — | — no latency data | No data |
NVIDIA A100 80GB no rental price | — | — | — | — no latency data | No data |
NVIDIA Jetson AGX Thor no rental price | — | — | — | — no latency data | No data |
NVIDIA H100 80GB no rental price | — | — | — | — no latency data | No data |
NVIDIA A800 80GB no rental price | — | — | — | — no latency data | No data |
No latency measurements for this model yet · Data comes from literature, maintainer runs, or crowd CLI uploads. None exist so far.
Compatibility
Colored only where we hold evidence. An amber dot means the vendor declares this form as a target; that is a claim, not a verified compatibility.
| Unitree G1 | Unitree H1 | AgiBot Genie G1 | AgileX Piper arm | SO-101 | Franka FR3 / Panda | Fourier GR-2 | WidowX 250 S | Google Robot（Everyday Robots） | ALOHA 2 / Cobot Magic (dual-arm) | Galaxea R1 Pro | Franka Emika Panda | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Octo |
Legend Officially supportedCommunity reproducedFine-tunable in theoryUnsupportedUnknown Vendor-declared target form 4 / 12 cells with a definite status
Reproduction records
Did this model run on your robot? Report it with a link (log, video or repo). Records are reviewed by the maintainers; a verified “works” or “works after fine-tuning” upgrades the matrix cell to “community verified” with your link as evidence. Anonymous is fine; nothing but the fields below is stored.
No verified record yet for this model.
Clear direction, grounded judgement.