Liquid AILanguage modelOpen weights
LFM3-Omni-2B
LFM3-Omni-2B has no overall index value yet because there are too few independent benchmark results.
Summary
LFM3-Omni-2B combines text, image, video, and audio in one local model. It processes video with its soundtrack, accepts 16 kHz audio, and returns text, 24 kHz speech, or both. The card lists 32K context tokens and a default of 2 fps with 4 to 384 video frames.
Liquid AI publishes separate tables for audiovisual questions, vision-only tasks, and audio. They include ten audiovisual or spoken-visual tests, 17 vision benchmarks, and six audio measures. Vision scores use 0 to 100 except OCRBench, which uses a 1,000-point scale. Audio keeps its own units.
All benchmark results
Every published result with its source, test setup and comparison models.
Show all 33 published benchmark results
Page 1 of 3
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
Daily-Omni (Show measurement, test conditions, and source)
- Source value
- 70.09
- Score
- 70.09
- Metric
- percent
- Unit
- %
- Category
- Video + audio understanding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
WorldSense (Show measurement, test conditions, and source)
- Source value
- 47.1
- Score
- 47.1
- Metric
- percent
- Unit
- %
- Category
- Video + audio understanding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
AV-SpeakerBench (Show measurement, test conditions, and source)
- Source value
- 46.79
- Score
- 46.79
- Metric
- percent
- Unit
- %
- Category
- Video + audio understanding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
OmniBench (overall) (Show measurement, test conditions, and source)
- Source value
- 46.32
- Score
- 46.32
- Metric
- percent
- Unit
- %
- Category
- Image + audio understanding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
OmniBench (speech) (Show measurement, test conditions, and source)
- Source value
- 48.25
- Score
- 48.25
- Metric
- percent
- Unit
- %
- Category
- Image + audio understanding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
RealWorldQA (Show measurement, test conditions, and source)
- Source value
- 64.68
- Score
- 64.68
- Metric
- percent
- Unit
- %
- Category
- Spoken visual QA
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
MMStar (Show measurement, test conditions, and source)
- Source value
- 51.79
- Score
- 51.79
- Metric
- percent
- Unit
- %
- Category
- Spoken visual QA
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
POPE (Show measurement, test conditions, and source)
- Source value
- 83.78
- Score
- 83.78
- Metric
- percent
- Unit
- %
- Category
- Spoken visual QA
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
ERQA (Show measurement, test conditions, and source)
- Source value
- 39.43
- Score
- 39.43
- Metric
- percent
- Unit
- %
- Category
- Spoken visual QA
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
EmbSpatial (Show measurement, test conditions, and source)
- Source value
- 67.59
- Score
- 67.59
- Metric
- percent
- Unit
- %
- Category
- Spoken visual QA
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Video and audio or spoken visual QA table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
MMStar (Show measurement, test conditions, and source)
- Source value
- 55.67
- Score
- 55.67
- Metric
- percent
- Unit
- %
- Category
- General
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Vision only table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
MMMU (val) (Show measurement, test conditions, and source)
- Source value
- 42.44
- Score
- 42.44
- Metric
- percent
- Unit
- %
- Category
- General
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- LiquidAI/LFM3-Omni-2B official model card at revision 5797d07c125503bd4f4c0e7f915ae2d3d703e8de
- Status
- active
- Retrieved at
- 2026-10-29
- Methodology
- Vendor-reported result. Liquid AI LFM3-Omni-2B model-card Vision only table. Liquid AI model-card evaluation with model-specific templates, prompts, image and video preprocessing, and greedy text decoding. Vision and audiovisual scores use 0 to 100 except OCRBench out of 1,000; audio metrics retain their own units.
- Context
- Manufacturer-reported; these sources do not provide an independent reproduction of this result.
Page 1 of 3
All specifications
Architecture
- Architecture
- LFM3 backbone with the LFM3 vision encoder, causal audio encoder, and LiquidCodec audio tokenizer.
- Source: Architecture
Input and output
- Input modalities
- Text, images, video with or without soundtrack, and 16 kHz audio
- Source: Input modalities
Licence and access
- Access
- Open weights, local deployment, and the API and Playground linked by the publisher
- Source: Access
Further details
- Context and video
- 32K context tokens. Video is processed at a default 2 fps with 4 to 384 frames. The soundtrack is processed in full.
- Source: Context and video
- Speech languages and locales
- 16 text languages and 18 listed speech locales for speech features.
- Source: Speech languages and locales
- Benchmark blocks
- The model card separates audiovisual and spoken visual questions from vision-only and audio tests. The model card publishes separate tables for these groups.
- Source: Benchmark blocks
Frequently asked questions about LFM3-Omni-2B
How large is the context window of LFM3-Omni-2B?
32,768 tokens. That is how much text LFM3-Omni-2B reads at once, about 50 A4 pages.
Page count as a rule of thumb from Artificial Analysis
DeepSeek-V3 0324164K
GLM-4.5131K
GPT OSS 120B131K
GPT OSS 20B131K
DeepSeek-V3.2128K
DeepSeek-R1128K
LFM3-Omni-2BRank 100 of 10033K
Can I run LFM3-Omni-2B locally?
Yes, the weights are open.
Stated by Liquid AI
- Open weights
- yes
- License
- LFM Open License v1.0. See the license text for its terms.
- Parameters
- 2.2B
- active per token
- n/a
Sources and data date
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
No active result from the reviewed external evaluators can currently be attributed to this model. Other source evidence remains available below.
Without vendor claims and historical or disputed results; those stay labelled in the single values.
| Type | Evidence and data date |
|---|---|
| Input | Documented modalities. Retrieved October 29, 2026. |
| Output | Documented modalities. Retrieved October 29, 2026. |
| Research status | Research date October 29, 2026. 62 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not published | Capability radar raw values Note The additional radar chart publishes no numeric axis values. Its qualitative bars were not converted into invented scores. |
| Not published | ASR sample counts Note The card names the ESB/Open-ASR suite and nine FLEURS languages but gives no per-language audio-file counts. |
| Not found in the inspected sources | Independent evaluation Note The reviewed manufacturer sources do not link an independent reproduction of the LFM3-Omni results. |
| Not found in the inspected sources | Technical report or research paper Note Reviewed publisher sources do not link a separate technical report or research paper for LFM3-Omni-2B. |
| Not found in the inspected sources | Vals rows in captured payloads Note Nine complete Vals payloads dated 28 October were checked; no exact model ID or unambiguous model label. Limited to these saved benchmarks. |
| Unresolved | Uncaptured Arena Code/Speech leaderboards Note Text, Agent including Code/Chat/Work, and twelve WebDev views were captured. A separate general Code and Arena Speech leaderboard were not saved as complete tables. |
| Unresolved | Artificial Analysis catalog reconciliation Note The full current AA model and benchmark catalog was not re-reconciled for this profile in this sidecar. This sidecar makes no AA absence claim; native AA imports remain authoritative. |
| Not found in the inspected sources | VoiceArena STT exact profile row Note The saved VoiceArena speech-to-text results do not contain this exact model identity. This is limited to that capture and does not rule out future rows or other speech evaluators. |
| Additional source | Liquid AI LFM3 release announcement (retrieved October 29, 2026) · Liquid AI LFM3 release announcement · Editorial description reviewed October 29, 2026 |
| Additional source | Liquid AI model documentation (retrieved October 29, 2026) · Liquid AI Omni documentation · Editorial description reviewed October 29, 2026 |
| Additional source | Liquid AI LFM3 Hugging Face collection (retrieved October 29, 2026) |
| Additional source | Hugging Face model metadata (retrieved October 29, 2026) |
| Additional source | Model metadata (retrieved October 29, 2026)123 |