LLM Compare
Head to head

LFM3-VL-2B vs ERNIE 5.5 Flash Preview

How LFM3-VL-2B and ERNIE 5.5 Flash Preview stack up across benchmarks, pricing, speed, and the workloads that matter.

LFM3-VL-2B

Liquid AI · Open weights

Blended pricen/aper 1M tokens
Contextn/atokens
Speedn/atok/s
Benchmarks41results
VS

ERNIE 5.5 Flash Preview

Baidu · Proprietary

Blended price$0.37per 1M tokens
Context1.05Mtokens
Speedn/atok/s
Benchmarks13results

On math, LFM3-VL-2B leads by about 24 points in our normalized benchmark average.

The table below breaks down every workload axis where both models have published results. Each score is a normalized 0 to 1 average across the benchmarks tagged with that category. A gap of a few points is noise; a gap of ten or more is a real difference in capability. Where one model has not reported a result, the cell shows n/a and the missing benchmark does not drag its average down.

Benchmark scores by workload

WorkloadLFM3-VL-2BERNIE 5.5 Flash PreviewEdge
Reasoning 66% 50% +16 pts
Coding n/a 58% n/a
Agents & tool use n/a 53% n/a
Math 71% 46% +24.5 pts
Vision & multimodal 70% n/a n/a
Long context 41% n/a n/a
Writing n/a 36% n/a
Knowledge & factuality 57% 44% +12.2 pts

The headline benchmarks below are the most widely cited individual tests. GPQA Diamond measures graduate-level reasoning in physics, chemistry, and biology. SWE-Bench Verified tests whether a model can fix real GitHub issues. MMMU-Pro covers college-level multimodal understanding across six disciplines. AIME and FrontierMath push competitive and research math. BrowseComp measures web research ability.

Headline benchmarks

BenchmarkLFM3-VL-2BERNIE 5.5 Flash Preview
MMMU-Pro 46% n/a
BrowseComp n/a 87%

Benchmarks tell you what a model can do in a controlled setting. They do not tell you whether it will work on your specific task, with your specific data, at your specific scale. The recommendations below map each common workload to whichever of these two models scores higher on the relevant axis. Use them as a starting point, not a final answer.

Which should you choose?

Hard reasoning

Multi-step analysis, research, and problems that need sustained thought.

LFM3-VL-2B

Coding & software

Writing, reviewing, and debugging code across a real codebase.

ERNIE 5.5 Flash Preview

Agents & tool use

Long-running agents that call tools, browse, and act on their own.

ERNIE 5.5 Flash Preview

Math & science

Formal math, competitive problems, and quantitative science.

LFM3-VL-2B

Vision & documents

Reading images, screenshots, charts, and dense documents.

LFM3-VL-2B

Long documents

Whole codebases, books, and transcripts that fill the context window.

LFM3-VL-2B

Writing & drafting

Copy, emails, and long-form drafting where tone matters.

ERNIE 5.5 Flash Preview

Scores are normalized from public benchmarks published by llm-stats.com and averaged per workload. Pricing is the blended input/output cost per million tokens at an 8:1 mix. Refreshes daily. How this works.