MolmoAct 2 vs Ternary Bonsai 27B
At a Glance
| Compare | MolmoAct 2Ai2 | Ternary Bonsai 27BPrismML |
|---|---|---|
| Pricing and Limits | ||
| Context windowMaximum documented tokens | Not reported | 262K |
| Model facts checked | Aug 29, 2026View model evidence → | Sep 18, 2026View model evidence → |
Available Benchmarks
Side-by-Side Facts
| Field | MolmoAct 2 | Ternary Bonsai 27B |
|---|---|---|
| Developer | Ai2 | PrismML |
| Family | MolmoAct 2 | Bonsai 27b |
| Model | MolmoAct 2 | Ternary Bonsai 27B |
| Version | 2 | Ternary Bonsai 27B |
| Lifecycle | active | active |
| Released | 2026-05-05 | 2026-07-04 |
| Knowledge cutoff | Unknown | Unknown |
| Input modalities | Text, Image, Robot state | Text, Image |
| Output modalities | Robot action | Text |
| Context window | Unknown | 262K |
| Total parameters | 5B | 27B |
| Active parameters | Unknown | Unknown |
| License | apache-2.0 | apache-2.0 |
| Open weights | Yes | Yes |
| API available | No | Yes |
| Self-hostable | Yes | Yes |
| Provider access | Unknown | Together Ai (Standard) |
| Capabilities | action-reasoning, bimanual-manipulation, depth-reasoning, fine-tuning | chat, generation, reasoning, tools, vision |
| Base model | Unknown | Qwen3.6 27B |
| Effective bit width | Unknown | 1.58 bits per weight |
| Language model size | Unknown | 6.66 GiB |
| Weight format | Unknown | Ternary Q2_0 |
| Robotics model type | Action reasoning model | Unknown |
| Action representation | Flow-matching continuous action expert | Unknown |
| Control architecture | Molmo 2-ER backbone with KV-cache bridge and action expert | Unknown |
| Inference location | Flexible | Unknown |
| Native control rate (Hz) | Unknown | Unknown |
| Supported embodiments | SO-100, SO-101, Franka, WidowX, bimanual YAM | Unknown |
| Training data | Open bimanual YAM, SO-100/SO-101, DROID, BC-Z, Fractal, Bridge, and prior MolmoAct data described by Ai2. | Unknown |
MolmoAct 2 Capabilities
Ternary Bonsai 27B Capabilities
Primary Evidence
Sources and Freshness
Questions
MolmoAct 2 vs Ternary Bonsai 27B FAQs
Is MolmoAct 2 or Ternary Bonsai 27B better for coding?+
This comparison does not currently contain a protocol-matched coding benchmark for both MolmoAct 2 and Ternary Bonsai 27B, so Model Markets cannot name a coding leader from pricing, context size, or capability labels alone.
Which is cheaper, MolmoAct 2 or Ternary Bonsai 27B?+
Neither model has a directly sourced input price in this comparison. Neither model has a directly sourced output price in this comparison.
Which has a larger context window, MolmoAct 2 or Ternary Bonsai 27B?+
Neither model has a larger sourced context window in this comparison. MolmoAct 2 is — and Ternary Bonsai 27B is 262K.
Which performs better in benchmarks, MolmoAct 2 or Ternary Bonsai 27B?+
There is no overall benchmark winner: At least two independently verified, protocol-matched benchmarks are required for an overall winner.
Can MolmoAct 2 or Ternary Bonsai 27B be self-hosted?+
Both models have the same recorded self-hosting status: supported. MolmoAct 2 is open weight; Ternary Bonsai 27B is open weight.
Can MolmoAct 2 and Ternary Bonsai 27B understand images?+
MolmoAct 2 is documented with image input; Ternary Bonsai 27B is documented with image input. This reflects supported input modalities, not vision quality.
Which can generate longer answers, MolmoAct 2 or Ternary Bonsai 27B?+
Neither has a larger sourced maximum output. MolmoAct 2 is — and Ternary Bonsai 27B is —.
Do MolmoAct 2 and Ternary Bonsai 27B support reasoning and tool use?+
MolmoAct 2: reasoning and image input. Ternary Bonsai 27B: reasoning, tool calling, and image input. Feature support does not establish relative quality.
Which is available from more inference providers, MolmoAct 2 or Ternary Bonsai 27B?+
MolmoAct 2 has 0 sourced provider routes; Ternary Bonsai 27B has 1, so Ternary Bonsai 27B has broader tracked availability.
Which offers better value, MolmoAct 2 or Ternary Bonsai 27B?+
There is no universal value winner. Compare the input and output prices above with the matched benchmark result for your workload: cheaper tokens can be offset by different quality, token usage, latency, or provider availability.