# LFM2.5-24B-A2B vs LFM3-1.7B

> Side-by-side benchmark comparison for LFM2.5-24B-A2B and LFM3-1.7B across agentic, coding, multimodal, knowledge, reasoning, multilingual, and math tasks.

- Canonical page: https://benchlm.ai/compare/lfm2-5-24b-a2b-vs-lfm3-1-7b
- Last updated: October 27, 2026

- Shared sourced benchmarks: 11
- HTML indexing: indexable
- Ranking lane: BenchAlign v5.8

## Quick Verdict

LFM2.5-24B-A2B leads by point estimate. Conditional score ranges do not establish rank confidence. Choose using the category evidence and your own workload tests.

## Summary

- LFM2.5-24B-A2B has the higher overall point estimate 36.99 to 22.99. Conditional score ranges do not establish rank confidence.
- The clearest category separation is in agentic, where the averages are 30.5 for LFM2.5-24B-A2B and 14.4 for LFM3-1.7B.
- The biggest single benchmark swing is τ³-bench results in Agentic, with scores of 8.0% and 68.0%.
- LFM3-1.7B is the cheaper option on output tokens, which matters if you expect large responses or heavy interactive use.
- LFM3-1.7B also has the larger context window at 256K.

## Model Snapshot

| Property | LFM2.5-24B-A2B | LFM3-1.7B |
|----------|----------|----------|
| Creator | LiquidAI | LiquidAI |
| Type | Open Weight | Open Weight |
| Reasoning | Reasoning | Reasoning |
| Context | 131K | 256K |
| Overall Score | 36.99 | 22.99 |
| Benchmarks Covered | 13 | 24 |
| Pricing (input/output) | $0.10 / $1.00 | $0.00 / $0.00 |

## Category Breakdown

### Agentic

- Winner: Directional only
- LFM2.5-24B-A2B public-lane score: 30.5 (Estimated · #99/134)
- LFM3-1.7B public-lane score: 14.4 (Estimated · #123/134)

| Benchmark | LFM2.5-24B-A2B | LFM3-1.7B | Winner |
|-----------|-----------|-----------|--------|
| AutomationBench | 27.7% | 41.3% | LFM3-1.7B |
| BFCL v4 | 65.0% | 71.4% | LFM3-1.7B |
| τ³-bench results | 8.0% | 68.0% | LFM3-1.7B |
| PinchBench | 83.2% | Coming soon | Coming soon |
| WebArena-Verified | 26.8% | 38.4% | LFM3-1.7B |
| MCP Atlas | Coming soon | 32.6% | Coming soon |
| ToolSandbox | Coming soon | 80.36% | Coming soon |
| Claw-Eval | Coming soon | 68.6% | Coming soon |
| HLE w/ tools | Coming soon | 13.9% | Coming soon |
| Terminal-Bench 2.1 | Coming soon | 12.9% | Coming soon |
| LiveClaw | Coming soon | 45.60% | Coming soon |

### Coding

- Winner: Not comparable
- LFM2.5-24B-A2B public-lane score: Coming soon
- LFM3-1.7B public-lane score: 13 (Estimated · #142/155)

| Benchmark | LFM2.5-24B-A2B | LFM3-1.7B | Winner |
|-----------|-----------|-----------|--------|
| SciCode | 32.7% | 28.5% | LFM2.5-24B-A2B |
| SWE-bench Verified | 51% | 43.4% | LFM2.5-24B-A2B |
| SWE Multilingual | Coming soon | 23% | Coming soon |
| Terminal-Bench 2.1 | Coming soon | 12.9% | Coming soon |
| LiveCodeBench v6 | Coming soon | 73.5% | Coming soon |
| LiveCodeBench Pro | Coming soon | 31.7% | Coming soon |

### Reasoning

- Winner: Not comparable
- LFM2.5-24B-A2B public-lane score: Coming soon
- LFM3-1.7B public-lane score: 61.3 (Unranked · 1 rankable row)

| Benchmark | LFM2.5-24B-A2B | LFM3-1.7B | Winner |
|-----------|-----------|-----------|--------|
| RULER (128K) | Coming soon | 65.56% | Coming soon |

### Knowledge

- Winner: Directional only
- LFM2.5-24B-A2B public-lane score: 39.4 (Estimated · #124/189)
- LFM3-1.7B public-lane score: 25.5 (Estimated · #175/189)

| Benchmark | LFM2.5-24B-A2B | LFM3-1.7B | Winner |
|-----------|-----------|-----------|--------|
| GPQA | 74.4% | 63.4% | LFM2.5-24B-A2B |
| GPQA-D | 74.4% | 63.4% | LFM2.5-24B-A2B |
| HLE | 7.4% | 7.2% | LFM2.5-24B-A2B |

### Instruction Following

- Winner: LFM2.5-24B-A2B
- LFM2.5-24B-A2B public-lane score: 75.6 (#69/156)
- LFM3-1.7B public-lane score: 71.7 (#76/156)

| Benchmark | LFM2.5-24B-A2B | LFM3-1.7B | Winner |
|-----------|-----------|-----------|--------|
| IFEval | 96.8% | 96.7% | LFM2.5-24B-A2B |
| IFBench | 74.7% | 73.0% | LFM2.5-24B-A2B |
| Multi-IF | Coming soon | 89.10% | Coming soon |
| IFStruct | Coming soon | 93.72% | Coming soon |

### Mathematics

- Winner: Not comparable
- LFM2.5-24B-A2B public-lane score: Coming soon
- LFM3-1.7B public-lane score: Coming soon

| Benchmark | LFM2.5-24B-A2B | LFM3-1.7B | Winner |
|-----------|-----------|-----------|--------|
| MATH-500 | 96.0% | Coming soon | Coming soon |

## FAQ

### Which is better overall, LFM2.5-24B-A2B or LFM3-1.7B?

LFM2.5-24B-A2B has the higher overall point estimate. Conditional score ranges do not establish rank confidence.

### Where is the biggest gap between LFM2.5-24B-A2B and LFM3-1.7B?

The widest category gap is in agentic, where the averages are 30.5 for LFM2.5-24B-A2B and 14.4 for LFM3-1.7B.

### How many benchmarks does LFM2.5-24B-A2B cover on BenchLM?

LFM2.5-24B-A2B currently has 13 sourced benchmark scores on BenchLM.

### How many benchmarks does LFM3-1.7B cover on BenchLM?

LFM3-1.7B currently has 24 sourced benchmark scores on BenchLM.

## Related Comparisons

- [LFM2.5-24B-A2B vs LFM3-1.7B-Base](/compare/lfm2-5-24b-a2b-vs-lfm3-1-7b-base)
- [LFM3-1.7B vs LFM3-1.7B-Base](/compare/lfm3-1-7b-vs-lfm3-1-7b-base)
- [LFM2.5-24B-A2B vs LFM3-Audio-2B](/compare/lfm2-5-24b-a2b-vs-lfm3-audio-2b)
- [LFM3-1.7B vs LFM3-Audio-2B](/compare/lfm3-1-7b-vs-lfm3-audio-2b)

## Explore More

- [LFM2.5-24B-A2B profile](/models/lfm2-5-24b-a2b)
- [LFM3-1.7B profile](/models/lfm3-1-7b)
- [Compare Pricing](/llm-pricing)
- [Alternative Finder](/tools/alternative-finder)
- [LLM Selector](/tools/llm-selector)
- [Overall Rankings](/best/overall)